NetBSD/usr.sbin/dumplfs/dumplfs.c

807 lines
23 KiB
C
Raw Normal View History

2005-08-19 06:07:18 +04:00
/* $NetBSD: dumplfs.c,v 1.29 2005/08/19 02:09:50 christos Exp $ */
/*-
* Copyright (c) 1991, 1993
* The Regents of the University of California. All rights reserved.
*
* 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. 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.
*/
1998-03-01 05:20:01 +03:00
#include <sys/cdefs.h>
#ifndef lint
1998-03-01 05:20:01 +03:00
__COPYRIGHT(
"@(#) Copyright (c) 1991, 1993\n\
1998-03-01 05:20:01 +03:00
The Regents of the University of California. All rights reserved.\n");
#endif /* not lint */
#ifndef lint
#if 0
1998-03-01 05:20:01 +03:00
static char sccsid[] = "@(#)dumplfs.c 8.5 (Berkeley) 5/24/95";
#else
2005-08-19 06:07:18 +04:00
__RCSID("$NetBSD: dumplfs.c,v 1.29 2005/08/19 02:09:50 christos Exp $");
#endif
#endif /* not lint */
#include <sys/param.h>
#include <sys/ucred.h>
#include <sys/mount.h>
#include <sys/time.h>
#include <ufs/ufs/dinode.h>
#include <ufs/lfs/lfs.h>
1994-09-23 06:18:30 +04:00
#include <err.h>
#include <errno.h>
#include <fcntl.h>
#include <fstab.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
1994-09-23 06:18:30 +04:00
#include <unistd.h>
#include "extern.h"
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
static void addseg(char *);
static void dump_cleaner_info(struct lfs *, void *);
static void dump_dinode(struct ufs1_dinode *);
static void dump_ifile(int, struct lfs *, int, int, daddr_t);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
static int dump_ipage_ifile(struct lfs *, int, char *, int);
static int dump_ipage_segusage(struct lfs *, int, char *, int);
static void dump_segment(int, int, daddr_t, struct lfs *, int);
static int dump_sum(int, struct lfs *, SEGSUM *, int, daddr_t);
static void dump_super(struct lfs *);
static void usage(void);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
extern u_long cksum(void *, size_t);
1998-03-01 05:20:01 +03:00
typedef struct seglist SEGLIST;
struct seglist {
SEGLIST *next;
int num;
};
SEGLIST *seglist;
char *special;
/* Segment Usage formats */
#define print_suheader \
(void)printf("segnum\tflags\tnbytes\tninos\tnsums\tlastmod\n")
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
#define print_suentry(i, sp, fs) \
(void)printf("%d\t%c%c%c\t%d\t%d\t%d\t%s", i, \
(((sp)->su_flags & SEGUSE_ACTIVE) ? 'A' : ' '), \
(((sp)->su_flags & SEGUSE_DIRTY) ? 'D' : 'C'), \
(((sp)->su_flags & SEGUSE_SUPERBLOCK) ? 'S' : ' '), \
(sp)->su_nbytes, (sp)->su_ninos, (sp)->su_nsums, \
((fs)->lfs_version == 1 ? ctime((time_t *)&(sp)->su_olastmod) : \
ctime((time_t *)&(sp)->su_lastmod)))
/* Ifile formats */
#define print_iheader \
(void)printf("inum\tstatus\tversion\tdaddr\t\tfreeptr\n")
#define print_ientry(i, ip) \
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if ((ip)->if_daddr == LFS_UNUSED_DADDR) \
2005-08-19 06:07:18 +04:00
(void)printf("%d\tFREE\t%d\t \t\t%llu\n", \
i, (ip)->if_version, \
(unsigned long long)(ip)->if_nextfree); \
else \
(void)printf("%d\tINUSE\t%d\t%8X \n", \
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
i, (ip)->if_version, (ip)->if_daddr)
#define fsbtobyte(fs, b) fsbtob((fs), (off_t)((b)))
int datasum_check = 0;
int
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
main(int argc, char **argv)
{
struct lfs lfs_sb1, lfs_sb2, *lfs_master;
daddr_t seg_addr, idaddr, sbdaddr;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
int ch, do_allsb, do_ientries, do_segentries, fd, segnum;
do_allsb = 0;
do_ientries = 0;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
do_segentries = 0;
idaddr = 0x0;
sbdaddr = 0x0;
while ((ch = getopt(argc, argv, "ab:diI:Ss:")) != -1)
switch(ch) {
case 'a': /* Dump all superblocks */
do_allsb = 1;
break;
case 'b': /* Use this superblock */
sbdaddr = strtol(optarg, NULL, 0);
break;
case 'd':
datasum_check = 1;
break;
case 'i': /* Dump ifile entries */
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
do_ientries = !do_ientries;
break;
case 'I': /* Use this ifile inode */
idaddr = strtol(optarg, NULL, 0);
break;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
case 'S':
do_segentries = !do_segentries;
break;
case 's': /* Dump out these segments */
addseg(optarg);
break;
default:
usage();
}
argc -= optind;
argv += optind;
if (argc != 1)
usage();
special = argv[0];
if ((fd = open(special, O_RDONLY, 0)) < 0)
1994-09-23 06:18:30 +04:00
err(1, "%s", special);
if (sbdaddr == 0x0) {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
/* Read the proto-superblock */
get(fd, LFS_LABELPAD, &(lfs_sb1.lfs_dlfs), sizeof(struct dlfs));
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
/* If that wasn't the real first sb, get the real first sb */
if (lfs_sb1.lfs_version > 1 &&
lfs_sb1.lfs_sboffs[0] > btofsb(&lfs_sb1, LFS_LABELPAD))
get(fd, fsbtob(&lfs_sb1, lfs_sb1.lfs_sboffs[0]),
&(lfs_sb1.lfs_dlfs), sizeof(struct dlfs));
/*
* Read the second superblock and figure out which check point is
* most up to date.
*/
get(fd,
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
fsbtobyte(&lfs_sb1, lfs_sb1.lfs_sboffs[1]),
&(lfs_sb2.lfs_dlfs), sizeof(struct dlfs));
lfs_master = &lfs_sb1;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (lfs_sb1.lfs_version > 1) {
if (lfs_sb1.lfs_serial > lfs_sb2.lfs_serial) {
lfs_master = &lfs_sb2;
sbdaddr = lfs_sb1.lfs_sboffs[1];
} else
sbdaddr = lfs_sb1.lfs_sboffs[0];
} else {
if (lfs_sb1.lfs_otstamp > lfs_sb2.lfs_otstamp) {
lfs_master = &lfs_sb2;
sbdaddr = lfs_sb1.lfs_sboffs[1];
} else
sbdaddr = lfs_sb1.lfs_sboffs[0];
}
} else {
/* Read the first superblock */
get(fd, dbtob((off_t)sbdaddr), &(lfs_sb1.lfs_dlfs),
sizeof(struct dlfs));
lfs_master = &lfs_sb1;
}
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
/* Compatibility */
if (lfs_master->lfs_version == 1) {
lfs_master->lfs_sumsize = LFS_V1_SUMMARY_SIZE;
lfs_master->lfs_ibsize = lfs_master->lfs_bsize;
lfs_master->lfs_start = lfs_master->lfs_sboffs[0];
lfs_master->lfs_tstamp = lfs_master->lfs_otstamp;
lfs_master->lfs_fsbtodb = 0;
}
(void)printf("Master Superblock at 0x%llx:\n", (long long)sbdaddr);
dump_super(lfs_master);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_ifile(fd, lfs_master, do_ientries, do_segentries, idaddr);
if (seglist != NULL)
for (; seglist != NULL; seglist = seglist->next) {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
seg_addr = sntod(lfs_master, seglist->num);
dump_segment(fd, seglist->num, seg_addr, lfs_master,
do_allsb);
}
else
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
for (segnum = 0, seg_addr = sntod(lfs_master, 0);
segnum < lfs_master->lfs_nseg;
segnum++, seg_addr = sntod(lfs_master, segnum))
dump_segment(fd, segnum, seg_addr, lfs_master,
do_allsb);
(void)close(fd);
exit(0);
}
/*
* We are reading all the blocks of an inode and dumping out the ifile table.
* This code could be tighter, but this is a first pass at getting the stuff
* printed out rather than making this code incredibly efficient.
*/
static void
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_ifile(int fd, struct lfs *lfsp, int do_ientries, int do_segentries, daddr_t addr)
{
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
char *ipage;
struct ufs1_dinode *dip, *dpage;
/* XXX ondisk32 */
int32_t *addrp, *dindir, *iaddrp, *indir;
int block_limit, i, inum, j, nblocks, psize;
psize = lfsp->lfs_bsize;
if (!addr)
addr = lfsp->lfs_idaddr;
if (!(dpage = malloc(psize)))
1994-09-23 06:18:30 +04:00
err(1, "malloc");
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, addr), dpage, psize);
for (dip = dpage + INOPB(lfsp) - 1; dip >= dpage; --dip)
if (dip->di_inumber == LFS_IFILE_INUM)
break;
if (dip < dpage)
errx(1, "unable to locate ifile inode at disk address 0x%llx",
(long long)addr);
(void)printf("\nIFILE inode\n");
dump_dinode(dip);
(void)printf("\nIFILE contents\n");
nblocks = dip->di_size >> lfsp->lfs_bshift;
block_limit = MIN(nblocks, NDADDR);
/* Get the direct block */
if ((ipage = malloc(psize)) == NULL)
1994-09-23 06:18:30 +04:00
err(1, "malloc");
for (inum = 0, addrp = dip->di_db, i = 0; i < block_limit;
i++, addrp++) {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, *addrp), ipage, psize);
if (i < lfsp->lfs_cleansz) {
dump_cleaner_info(lfsp, ipage);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (do_segentries)
print_suheader;
continue;
}
if (i < (lfsp->lfs_segtabsz + lfsp->lfs_cleansz)) {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (do_segentries)
inum = dump_ipage_segusage(lfsp, inum, ipage,
lfsp->lfs_sepb);
else
inum = (i < lfsp->lfs_segtabsz + lfsp->lfs_cleansz - 1);
1998-08-27 22:03:42 +04:00
if (!inum) {
if(!do_ientries)
goto e0;
else
print_iheader;
1998-08-27 22:03:42 +04:00
}
} else
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
inum = dump_ipage_ifile(lfsp, inum, ipage, lfsp->lfs_ifpb);
}
if (nblocks <= NDADDR)
goto e0;
/* Dump out blocks off of single indirect block */
if (!(indir = malloc(psize)))
1994-09-23 06:18:30 +04:00
err(1, "malloc");
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, dip->di_ib[0]), indir, psize);
block_limit = MIN(i + lfsp->lfs_nindir, nblocks);
for (addrp = indir; i < block_limit; i++, addrp++) {
if (*addrp == LFS_UNUSED_DADDR)
break;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, *addrp), ipage, psize);
if (i < lfsp->lfs_cleansz) {
dump_cleaner_info(lfsp, ipage);
continue;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
}
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (i < lfsp->lfs_segtabsz + lfsp->lfs_cleansz) {
if (do_segentries)
inum = dump_ipage_segusage(lfsp, inum, ipage,
lfsp->lfs_sepb);
else
inum = (i < lfsp->lfs_segtabsz + lfsp->lfs_cleansz - 1);
1998-08-30 05:40:15 +04:00
if (!inum) {
if(!do_ientries)
goto e1;
else
print_iheader;
1998-08-30 05:40:15 +04:00
}
} else
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
inum = dump_ipage_ifile(lfsp, inum, ipage, lfsp->lfs_ifpb);
}
if (nblocks <= lfsp->lfs_nindir * lfsp->lfs_ifpb)
goto e1;
/* Get the double indirect block */
if (!(dindir = malloc(psize)))
1994-09-23 06:18:30 +04:00
err(1, "malloc");
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, dip->di_ib[1]), dindir, psize);
for (iaddrp = dindir, j = 0; j < lfsp->lfs_nindir; j++, iaddrp++) {
if (*iaddrp == LFS_UNUSED_DADDR)
break;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, *iaddrp), indir, psize);
block_limit = MIN(i + lfsp->lfs_nindir, nblocks);
for (addrp = indir; i < block_limit; i++, addrp++) {
if (*addrp == LFS_UNUSED_DADDR)
break;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, *addrp), ipage, psize);
if (i < lfsp->lfs_cleansz) {
dump_cleaner_info(lfsp, ipage);
continue;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
}
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (i < lfsp->lfs_segtabsz + lfsp->lfs_cleansz) {
if (do_segentries)
inum = dump_ipage_segusage(lfsp,
inum, ipage, lfsp->lfs_sepb);
else
inum = (i < lfsp->lfs_segtabsz +
lfsp->lfs_cleansz - 1);
1998-08-27 22:03:42 +04:00
if (!inum) {
if(!do_ientries)
goto e2;
else
print_iheader;
1998-08-27 22:03:42 +04:00
}
} else
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
inum = dump_ipage_ifile(lfsp, inum,
ipage, lfsp->lfs_ifpb);
}
}
e2: free(dindir);
e1: free(indir);
e0: free(dpage);
free(ipage);
}
static int
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_ipage_ifile(struct lfs *lfsp, int i, char *pp, int tot)
{
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
char *ip;
int cnt, max, entsize;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (lfsp->lfs_version == 1)
entsize = sizeof(IFILE_V1);
else
entsize = sizeof(IFILE);
max = i + tot;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
for (ip = pp, cnt = i; cnt < max; cnt++, ip += entsize)
print_ientry(cnt, (IFILE *)ip);
return (max);
}
static int
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_ipage_segusage(struct lfs *lfsp, int i, char *pp, int tot)
{
SEGUSE *sp;
int cnt, max;
struct seglist *slp;
max = i + tot;
for (sp = (SEGUSE *)pp, cnt = i;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
cnt < lfsp->lfs_nseg && cnt < max; cnt++) {
if (seglist == NULL)
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
print_suentry(cnt, sp, lfsp);
else {
for (slp = seglist; slp != NULL; slp = slp->next)
if (cnt == slp->num) {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
print_suentry(cnt, sp, lfsp);
break;
}
}
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (lfsp->lfs_version > 1)
++sp;
else
sp = (SEGUSE *)((SEGUSE_V1 *)sp + 1);
}
if (max >= lfsp->lfs_nseg)
return (0);
else
return (max);
}
static void
dump_dinode(struct ufs1_dinode *dip)
{
int i;
1995-12-15 01:30:45 +03:00
time_t at, mt, ct;
at = dip->di_atime;
mt = dip->di_mtime;
ct = dip->di_ctime;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
(void)printf(" %so%o\t%s%d\t%s%d\t%s%d\t%s%llu\n",
"mode ", dip->di_mode,
"nlink ", dip->di_nlink,
"uid ", dip->di_uid,
"gid ", dip->di_gid,
"size ", (long long)dip->di_size);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
(void)printf(" %s%s %s%s %s%s",
1995-12-15 01:30:45 +03:00
"atime ", ctime(&at),
"mtime ", ctime(&mt),
"ctime ", ctime(&ct));
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
(void)printf(" inum %d\n", dip->di_inumber);
(void)printf(" Direct Addresses\n");
for (i = 0; i < NDADDR; i++) {
(void)printf("\t0x%x", dip->di_db[i]);
if ((i % 6) == 5)
(void)printf("\n");
}
for (i = 0; i < NIADDR; i++)
(void)printf("\t0x%x", dip->di_ib[i]);
(void)printf("\n");
}
static int
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_sum(int fd, struct lfs *lfsp, SEGSUM *sp, int segnum, daddr_t addr)
{
FINFO *fp;
int32_t *dp, *idp;
int i, j, acc;
int ck;
int numbytes, numblocks;
char *datap;
struct ufs1_dinode *inop;
size_t el_size;
u_int32_t datasum;
char *buf;
1998-03-01 05:20:01 +03:00
if (sp->ss_magic != SS_MAGIC ||
sp->ss_sumsum != (ck = cksum(&sp->ss_datasum,
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
lfsp->lfs_sumsize - sizeof(sp->ss_sumsum)))) {
/* Don't print "corrupt" if we're just too close to the edge */
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (dtosn(lfsp, addr + fsbtodb(lfsp, 1)) ==
dtosn(lfsp, addr))
(void)printf("dumplfs: %s %d address 0x%llx\n",
"corrupt summary block; segment", segnum,
(long long)addr);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
return (0);
}
if (lfsp->lfs_version > 1 && sp->ss_ident != lfsp->lfs_ident) {
(void)printf("dumplfs: %s %d address 0x%llx\n",
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
"summary from a former life; segment", segnum,
(long long)addr);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
return (0);
}
(void)printf("Segment Summary Info at 0x%llx\n", (long long)addr);
(void)printf(" %s0x%x\t%s%d\t%s%d\t%s%c%c\n %s0x%x\t%s0x%x",
"next ", sp->ss_next,
"nfinfo ", sp->ss_nfinfo,
"ninos ", sp->ss_ninos,
"flags ", (sp->ss_flags & SS_DIROP) ? 'D' : '-',
(sp->ss_flags & SS_CONT) ? 'C' : '-',
"sumsum ", sp->ss_sumsum,
"datasum ", sp->ss_datasum );
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (lfsp->lfs_version == 1)
(void)printf("\tcreate %s\n", ctime((time_t *)&sp->ss_ident));
else {
(void)printf("\tcreate %s", ctime((time_t *)&sp->ss_create));
(void)printf(" roll_id %-8x", sp->ss_ident);
(void)printf(" serial %lld\n", (long long)sp->ss_serial);
}
/* Dump out inode disk addresses */
dp = (int32_t *)sp;
dp += lfsp->lfs_sumsize / sizeof(int32_t);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
inop = malloc(lfsp->lfs_bsize);
printf(" Inode addresses:");
1998-03-01 05:20:01 +03:00
numbytes = 0;
numblocks = 0;
for (dp--, i = 0; i < sp->ss_ninos; dp--) {
++numblocks;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
numbytes += lfsp->lfs_ibsize; /* add bytes for inode block */
printf("\t0x%x {", *dp);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, fsbtobyte(lfsp, *dp), inop, lfsp->lfs_ibsize);
for (j = 0; i < sp->ss_ninos && j < INOPB(lfsp); j++, i++) {
if (j > 0)
(void)printf(", ");
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
(void)printf("%dv%d", inop[j].di_inumber, inop[j].di_gen);
}
(void)printf("}");
if (((i/INOPB(lfsp)) % 4) == 3)
(void)printf("\n");
}
free(inop);
printf("\n");
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (lfsp->lfs_version == 1)
fp = (FINFO *)((SEGSUM_V1 *)sp + 1);
else
fp = (FINFO *)(sp + 1);
for (i = 0; i < sp->ss_nfinfo; i++) {
1998-03-01 05:20:01 +03:00
(void)printf(" FINFO for inode: %d version %d nblocks %d lastlength %d\n",
fp->fi_ino, fp->fi_version, fp->fi_nblocks,
fp->fi_lastlength);
dp = &(fp->fi_blocks[0]);
numblocks += fp->fi_nblocks;
for (j = 0; j < fp->fi_nblocks; j++, dp++) {
(void)printf("\t%d", *dp);
if ((j % 8) == 7)
(void)printf("\n");
1998-03-01 05:20:01 +03:00
if (j == fp->fi_nblocks - 1)
numbytes += fp->fi_lastlength;
else
numbytes += lfsp->lfs_bsize;
}
if ((j % 8) != 0)
(void)printf("\n");
fp = (FINFO *)dp;
}
if (datasum_check == 0)
return (numbytes);
/*
* Now that we know the number of blocks, run back through and
* compute the data checksum. (A bad data checksum is not enough
* to prevent us from continuing, but it odes merit a warning.)
*/
idp = (int32_t *)sp;
idp += lfsp->lfs_sumsize / sizeof(int32_t);
--idp;
if (lfsp->lfs_version == 1) {
fp = (FINFO *)((SEGSUM_V1 *)sp + 1);
el_size = sizeof(unsigned long);
} else {
fp = (FINFO *)(sp + 1);
el_size = sizeof(u_int32_t);
}
datap = (char *)malloc(el_size * numblocks);
memset(datap, 0, el_size * numblocks);
acc = 0;
addr += btofsb(lfsp, lfsp->lfs_sumsize);
buf = malloc(lfsp->lfs_bsize);
for (i = 0; i < sp->ss_nfinfo; i++) {
while (addr == *idp) {
get(fd, fsbtobyte(lfsp, addr), buf, lfsp->lfs_ibsize);
memcpy(datap + acc * el_size, buf, el_size);
addr += btofsb(lfsp, lfsp->lfs_ibsize);
--idp;
++acc;
}
for (j = 0; j < fp->fi_nblocks; j++) {
get(fd, fsbtobyte(lfsp, addr), buf, lfsp->lfs_fsize);
memcpy(datap + acc * el_size, buf, el_size);
if (j == fp->fi_nblocks - 1)
addr += btofsb(lfsp, fp->fi_lastlength);
else
addr += btofsb(lfsp, lfsp->lfs_bsize);
++acc;
}
fp = (FINFO *)&(fp->fi_blocks[fp->fi_nblocks]);
}
while (addr == *idp) {
get(fd, fsbtobyte(lfsp, addr), buf, lfsp->lfs_ibsize);
memcpy(datap + acc * el_size, buf, el_size);
addr += btofsb(lfsp, lfsp->lfs_ibsize);
--idp;
++acc;
}
free(buf);
if (acc != numblocks)
printf("** counted %d blocks but should have been %d\n",
acc, numblocks);
datasum = cksum(datap, numblocks * el_size);
if (datasum != sp->ss_datasum)
printf("** computed datasum 0x%lx does not match given datasum 0x%lx\n", (unsigned long)datasum, (unsigned long)sp->ss_datasum);
free(datap);
1998-03-01 05:20:01 +03:00
return (numbytes);
}
static void
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_segment(int fd, int segnum, daddr_t addr, struct lfs *lfsp, int dump_sb)
{
struct lfs lfs_sb, *sbp;
SEGSUM *sump;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
char *sumblock;
1998-03-01 05:20:01 +03:00
int did_one, nbytes, sb;
off_t sum_offset;
daddr_t new_addr;
(void)printf("\nSEGMENT %lld (Disk Address 0x%llx)\n",
(long long)dtosn(lfsp, addr), (long long)addr);
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
sum_offset = fsbtobyte(lfsp, addr);
sumblock = malloc(lfsp->lfs_sumsize);
if (lfsp->lfs_version > 1 && segnum == 0) {
if (fsbtob(lfsp, lfsp->lfs_start) < LFS_LABELPAD) {
/* First segment eats the disklabel */
sum_offset += fragroundup(lfsp, LFS_LABELPAD) -
fsbtob(lfsp, lfsp->lfs_start);
addr += btofsb(lfsp, fragroundup(lfsp, LFS_LABELPAD)) -
lfsp->lfs_start;
printf("Disklabel at 0x0\n");
}
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
}
sb = 0;
did_one = 0;
do {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
get(fd, sum_offset, sumblock, lfsp->lfs_sumsize);
sump = (SEGSUM *)sumblock;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if ((lfsp->lfs_version > 1 &&
sump->ss_ident != lfsp->lfs_ident) ||
sump->ss_sumsum != cksum (&sump->ss_datasum,
lfsp->lfs_sumsize - sizeof(sump->ss_sumsum))) {
sbp = (struct lfs *)sump;
1998-03-01 05:20:01 +03:00
if ((sb = (sbp->lfs_magic == LFS_MAGIC))) {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
printf("Superblock at 0x%x\n",
(unsigned)btofsb(lfsp, sum_offset));
if (dump_sb) {
get(fd, sum_offset, &(lfs_sb.lfs_dlfs),
sizeof(struct dlfs));
dump_super(&lfs_sb);
}
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (lfsp->lfs_version > 1)
sum_offset += fragroundup(lfsp, LFS_SBPAD);
else
sum_offset += LFS_SBPAD;
} else if (did_one)
break;
else {
printf("Segment at 0x%llx empty or corrupt\n",
(long long)addr);
break;
}
} else {
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
nbytes = dump_sum(fd, lfsp, sump, segnum,
btofsb(lfsp, sum_offset));
1998-03-01 05:20:01 +03:00
if (nbytes)
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
sum_offset += lfsp->lfs_sumsize + nbytes;
else
sum_offset = 0;
did_one = 1;
}
/* If the segment ends right on a boundary, it still ends */
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
new_addr = btofsb(lfsp, sum_offset);
/* printf("end daddr = 0x%lx\n", (long)new_addr); */
if (dtosn(lfsp, new_addr) != dtosn(lfsp, addr))
break;
} while (sum_offset);
return;
}
static void
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_super(struct lfs *lfsp)
{
int i;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
(void)printf(" %s0x%-8x %s0x%-8x %s%-10d\n",
"magic ", lfsp->lfs_magic,
"version ", lfsp->lfs_version,
"size ", lfsp->lfs_size);
(void)printf(" %s%-10d %s%-10d %s%-10d\n",
"ssize ", lfsp->lfs_ssize,
"dsize ", lfsp->lfs_dsize,
"bsize ", lfsp->lfs_bsize);
(void)printf(" %s%-10d %s%-10d %s%-10d\n",
"fsize ", lfsp->lfs_fsize,
"frag ", lfsp->lfs_frag,
"minfree ", lfsp->lfs_minfree);
(void)printf(" %s%-10d %s%-10d %s%-10d\n",
"inopb ", lfsp->lfs_inopb,
"ifpb ", lfsp->lfs_ifpb,
"nindir ", lfsp->lfs_nindir);
(void)printf(" %s%-10d %s%-10d %s%-10d\n",
"nseg ", lfsp->lfs_nseg,
"sepb ", lfsp->lfs_sepb,
"cleansz ", lfsp->lfs_cleansz);
(void)printf(" %s%-10d %s0x%-8x %s%-10d\n",
"segtabsz ", lfsp->lfs_segtabsz,
"segmask ", lfsp->lfs_segmask,
"segshift ", lfsp->lfs_segshift);
(void)printf(" %s0x%-8qx %s%-10d %s0x%-8qX\n",
"bmask ", (long long)lfsp->lfs_bmask,
"bshift ", lfsp->lfs_bshift,
"ffmask ", (long long)lfsp->lfs_ffmask);
(void)printf(" %s%-10d %s0x%-8qx %s%u\n",
"ffshift ", lfsp->lfs_ffshift,
"fbmask ", (long long)lfsp->lfs_fbmask,
"fbshift ", lfsp->lfs_fbshift);
(void)printf(" %s%-10d %s%-10d %s0x%-8x\n",
"sushift ", lfsp->lfs_sushift,
"fsbtodb ", lfsp->lfs_fsbtodb,
"cksum ", lfsp->lfs_cksum);
(void)printf(" %s%-10d %s%-10d %s%-10d\n",
"nclean ", lfsp->lfs_nclean,
"dmeta ", lfsp->lfs_dmeta,
"minfreeseg ", lfsp->lfs_minfreeseg);
(void)printf(" %s0x%-8x %s%-9d %s%-10d\n",
"roll_id ", lfsp->lfs_ident,
"interleave ", lfsp->lfs_interleave,
"sumsize ", lfsp->lfs_sumsize);
2003-02-21 07:01:07 +03:00
(void)printf(" %s%-10d %s0x%-8qx\n",
"seg0addr ", lfsp->lfs_start,
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
"maxfilesize ", (long long)lfsp->lfs_maxfilesize);
(void)printf(" Superblock disk addresses:\n ");
for (i = 0; i < LFS_MAXNUMSB; i++) {
(void)printf(" 0x%-8x", lfsp->lfs_sboffs[i]);
if (i == (LFS_MAXNUMSB >> 1))
(void)printf("\n ");
}
(void)printf("\n");
(void)printf(" Checkpoint Info\n");
(void)printf(" %s%-10d %s0x%-8x %s%-10d\n",
Add code to UBCify LFS. This is still behind "#ifdef LFS_UBC" for now (there are still some details to work out) but expect that to go away soon. To support these basic changes (creation of lfs_putpages, lfs_gop_write, mods to lfs_balloc) several other changes were made, to wit: * Create a writer daemon kernel thread whose purpose is to handle page writes for the pagedaemon, but which also takes over some of the functions of lfs_check(). This thread is started the first time an LFS is mounted. * Add a "flags" parameter to GOP_SIZE. Current values are GOP_SIZE_READ, meaning that the call should return the size of the in-core version of the file, and GOP_SIZE_WRITE, meaning that it should return the on-disk size. One of GOP_SIZE_READ or GOP_SIZE_WRITE must be specified. * Instead of using malloc(...M_WAITOK) for everything, reserve enough resources to get by and use malloc(...M_NOWAIT), using the reserves if necessary. Use the pool subsystem for structures small enough that this is feasible. This also obsoletes LFS_THROTTLE. And a few that are not strictly necessary: * Moves the LFS inode extensions off onto a separately allocated structure; getting closer to LFS as an LKM. "Welcome to 1.6O." * Unified GOP_ALLOC between FFS and LFS. * Update LFS copyright headers to correct values. * Actually cast to unsigned in lfs_shellsort, like the comment says. * Keep track of which segments were empty before the previous checkpoint; any segments that pass two checkpoints both dirty and empty can be summarily cleaned. Do this. Right now lfs_segclean still works, but this should be turned into an effectless compatibility syscall.
2003-02-18 02:48:08 +03:00
"freehd ", lfsp->lfs_freehd,
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
"idaddr ", lfsp->lfs_idaddr,
"ifile ", lfsp->lfs_ifile);
(void)printf(" %s%-10d %s%-10d %s%-10d\n",
"uinodes ", lfsp->lfs_uinodes,
"bfree ", lfsp->lfs_bfree,
"avail ", lfsp->lfs_avail);
(void)printf(" %s%-10d %s0x%-8x %s0x%-8x\n",
"nfiles ", lfsp->lfs_nfiles,
"lastseg ", lfsp->lfs_lastseg,
"nextseg ", lfsp->lfs_nextseg);
(void)printf(" %s0x%-8x %s0x%-8x %s%-10lld\n",
"curseg ", lfsp->lfs_curseg,
"offset ", lfsp->lfs_offset,
"serial ", (long long)lfsp->lfs_serial);
(void)printf(" tstamp %s", ctime((time_t *)&lfsp->lfs_tstamp));
}
static void
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
addseg(char *arg)
{
SEGLIST *p;
if ((p = malloc(sizeof(SEGLIST))) == NULL)
1994-09-23 06:18:30 +04:00
err(1, "malloc");
p->next = seglist;
p->num = atoi(arg);
seglist = p;
}
static void
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
dump_cleaner_info(struct lfs *lfsp, void *ipage)
{
CLEANERINFO *cip;
cip = (CLEANERINFO *)ipage;
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
if (lfsp->lfs_version > 1) {
(void)printf("free_head %d\n", cip->free_head);
(void)printf("free_tail %d\n", cip->free_tail);
}
(void)printf("clean\t%d\tdirty\t%d\n",
cip->clean, cip->dirty);
(void)printf("bfree\t%d\tavail\t%d\n\n",
cip->bfree, cip->avail);
}
static void
Merge the short-lived perseant-lfsv2 branch into the trunk. Kernels and tools understand both v1 and v2 filesystems; newfs_lfs generates v2 by default. Changes for the v2 layout include: - Segments of non-PO2 size and arbitrary block offset, so these can be matched to convenient physical characteristics of the partition (e.g., stripe or track size and offset). - Address by fragment instead of by disk sector, paving the way for non-512-byte-sector devices. In theory fragments can be as large as you like, though in reality they must be smaller than MAXBSIZE in size. - Use serial number and filesystem identifier to ensure that roll-forward doesn't get old data and think it's new. Roll-forward is enabled for v2 filesystems, though not for v1 filesystems by default. - The inode free list is now a tailq, paving the way for undelete (undelete is not yet implemented, but can be without further non-backwards-compatible changes to disk structures). - Inode atime information is kept in the Ifile, instead of on the inode; that is, the inode is never written *just* because atime was changed. Because of this the inodes remain near the file data on the disk, rather than wandering all over as the disk is read repeatedly. This speeds up repeated reads by a small but noticeable amount. Other changes of note include: - The ifile written by newfs_lfs can now be of arbitrary length, it is no longer restricted to a single indirect block. - Fixed an old bug where ctime was changed every time a vnode was created. I need to look more closely to make sure that the times are only updated during write(2) and friends, not after-the-fact during a segment write, and certainly not by the cleaner.
2001-07-14 00:30:18 +04:00
usage(void)
{
2003-04-16 13:25:04 +04:00
(void)fprintf(stderr, "usage: dumplfs [-adiS] [-b blkno] [-I blkno] [-s segno] filesys|device\n");
exit(1);
}