freebsd-nq/sys/ufs/ffs/ffs_vfsops.c

2230 lines
57 KiB
C
Raw Normal View History

/*-
1994-05-24 10:09:53 +00:00
* Copyright (c) 1989, 1991, 1993, 1994
* 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.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)ffs_vfsops.c 8.31 (Berkeley) 5/20/95
1994-05-24 10:09:53 +00:00
*/
2003-06-11 06:34:30 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
1996-01-05 18:31:58 +00:00
#include "opt_quota.h"
#include "opt_ufs.h"
#include "opt_ffs.h"
#include "opt_ddb.h"
1996-01-05 18:31:58 +00:00
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/namei.h>
#include <sys/priv.h>
1994-05-24 10:09:53 +00:00
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/vnode.h>
#include <sys/mount.h>
#include <sys/bio.h>
1994-05-24 10:09:53 +00:00
#include <sys/buf.h>
#include <sys/conf.h>
#include <sys/fcntl.h>
1994-05-24 10:09:53 +00:00
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <security/mac/mac_framework.h>
#include <ufs/ufs/extattr.h>
#include <ufs/ufs/gjournal.h>
1994-05-24 10:09:53 +00:00
#include <ufs/ufs/quota.h>
#include <ufs/ufs/ufsmount.h>
#include <ufs/ufs/inode.h>
#include <ufs/ufs/ufs_extern.h>
#include <ufs/ffs/fs.h>
#include <ufs/ffs/ffs_extern.h>
#include <vm/vm.h>
#include <vm/uma.h>
#include <vm/vm_page.h>
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
#include <geom/geom.h>
#include <geom/geom_vfs.h>
#include <ddb/ddb.h>
static uma_zone_t uma_inode, uma_ufs1, uma_ufs2;
static int ffs_reload(struct mount *, struct thread *);
static int ffs_mountfs(struct vnode *, struct mount *, struct thread *);
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
static void ffs_oldfscompat_read(struct fs *, struct ufsmount *,
ufs2_daddr_t);
static void ffs_ifree(struct ufsmount *ump, struct inode *ip);
static int ffs_sync_lazy(struct mount *mp);
static vfs_init_t ffs_init;
static vfs_uninit_t ffs_uninit;
static vfs_extattrctl_t ffs_extattrctl;
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
static vfs_cmount_t ffs_cmount;
static vfs_unmount_t ffs_unmount;
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
static vfs_mount_t ffs_mount;
static vfs_statfs_t ffs_statfs;
static vfs_fhtovp_t ffs_fhtovp;
static vfs_sync_t ffs_sync;
1994-05-24 10:09:53 +00:00
1998-02-09 06:11:36 +00:00
static struct vfsops ufs_vfsops = {
.vfs_extattrctl = ffs_extattrctl,
.vfs_fhtovp = ffs_fhtovp,
.vfs_init = ffs_init,
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
.vfs_mount = ffs_mount,
.vfs_cmount = ffs_cmount,
.vfs_quotactl = ufs_quotactl,
.vfs_root = ufs_root,
.vfs_statfs = ffs_statfs,
.vfs_sync = ffs_sync,
.vfs_uninit = ffs_uninit,
.vfs_unmount = ffs_unmount,
.vfs_vget = ffs_vget,
.vfs_susp_clean = process_deferred_inactive,
1994-05-24 10:09:53 +00:00
};
VFS_SET(ufs_vfsops, ufs, 0);
2006-07-09 14:11:09 +00:00
MODULE_VERSION(ufs, 1);
static b_strategy_t ffs_geom_strategy;
static b_write_t ffs_bufwrite;
static struct buf_ops ffs_ops = {
.bop_name = "FFS",
.bop_write = ffs_bufwrite,
.bop_strategy = ffs_geom_strategy,
.bop_sync = bufsync,
#ifdef NO_FFS_SNAPSHOT
.bop_bdflush = bufbdflush,
#else
.bop_bdflush = ffs_bdflush,
#endif
};
/*
* Note that userquota and groupquota options are not currently used
* by UFS/FFS code and generally mount(8) does not pass those options
* from userland, but they can be passed by loader(8) via
* vfs.root.mountfrom.options.
*/
static const char *ffs_opts[] = { "acls", "async", "noatime", "noclusterr",
"noclusterw", "noexec", "export", "force", "from", "groupquota",
"multilabel", "nfsv4acls", "fsckpid", "snapshot", "nosuid", "suiddir",
"nosymfollow", "sync", "union", "userquota", NULL };
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
static int
ffs_mount(struct mount *mp)
1994-05-24 10:09:53 +00:00
{
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
struct vnode *devvp;
struct thread *td;
struct ufsmount *ump = 0;
struct fs *fs;
pid_t fsckpid = 0;
int error, flags;
uint64_t mntorflags;
accmode_t accmode;
struct nameidata ndp;
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
char *fspec;
1995-05-30 08:16:23 +00:00
td = curthread;
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
if (vfs_filteropt(mp->mnt_optnew, ffs_opts))
return (EINVAL);
if (uma_inode == NULL) {
uma_inode = uma_zcreate("FFS inode",
sizeof(struct inode), NULL, NULL, NULL, NULL,
UMA_ALIGN_PTR, 0);
uma_ufs1 = uma_zcreate("FFS1 dinode",
sizeof(struct ufs1_dinode), NULL, NULL, NULL, NULL,
UMA_ALIGN_PTR, 0);
uma_ufs2 = uma_zcreate("FFS2 dinode",
sizeof(struct ufs2_dinode), NULL, NULL, NULL, NULL,
UMA_ALIGN_PTR, 0);
}
vfs_deleteopt(mp->mnt_optnew, "groupquota");
vfs_deleteopt(mp->mnt_optnew, "userquota");
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
fspec = vfs_getopts(mp->mnt_optnew, "from", &error);
if (error)
return (error);
1994-05-24 10:09:53 +00:00
mntorflags = 0;
if (vfs_getopt(mp->mnt_optnew, "acls", NULL, NULL) == 0)
mntorflags |= MNT_ACLS;
if (vfs_getopt(mp->mnt_optnew, "snapshot", NULL, NULL) == 0) {
mntorflags |= MNT_SNAPSHOT;
/*
* Once we have set the MNT_SNAPSHOT flag, do not
* persist "snapshot" in the options list.
*/
vfs_deleteopt(mp->mnt_optnew, "snapshot");
vfs_deleteopt(mp->mnt_opt, "snapshot");
}
if (vfs_getopt(mp->mnt_optnew, "fsckpid", NULL, NULL) == 0 &&
vfs_scanopt(mp->mnt_optnew, "fsckpid", "%d", &fsckpid) == 1) {
/*
* Once we have set the restricted PID, do not
* persist "fsckpid" in the options list.
*/
vfs_deleteopt(mp->mnt_optnew, "fsckpid");
vfs_deleteopt(mp->mnt_opt, "fsckpid");
if (mp->mnt_flag & MNT_UPDATE) {
if (VFSTOUFS(mp)->um_fs->fs_ronly == 0 &&
vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0) == 0) {
vfs_mount_error(mp,
"Checker enable: Must be read-only");
return (EINVAL);
}
} else if (vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0) == 0) {
vfs_mount_error(mp,
"Checker enable: Must be read-only");
return (EINVAL);
}
/* Set to -1 if we are done */
if (fsckpid == 0)
fsckpid = -1;
}
if (vfs_getopt(mp->mnt_optnew, "nfsv4acls", NULL, NULL) == 0) {
if (mntorflags & MNT_ACLS) {
vfs_mount_error(mp,
"\"acls\" and \"nfsv4acls\" options "
"are mutually exclusive");
return (EINVAL);
}
mntorflags |= MNT_NFS4ACLS;
}
MNT_ILOCK(mp);
2010-02-10 18:56:49 +00:00
mp->mnt_flag |= mntorflags;
MNT_IUNLOCK(mp);
1994-05-24 10:09:53 +00:00
/*
* If updating, check whether changing from read-only to
* read/write; if there is no device name, that's all we do.
*/
if (mp->mnt_flag & MNT_UPDATE) {
ump = VFSTOUFS(mp);
fs = ump->um_fs;
devvp = ump->um_devvp;
if (fsckpid == -1 && ump->um_fsckpid > 0) {
if ((error = ffs_flushfiles(mp, WRITECLOSE, td)) != 0 ||
(error = ffs_sbupdate(ump, MNT_WAIT, 0)) != 0)
return (error);
DROP_GIANT();
g_topology_lock();
/*
* Return to normal read-only mode.
*/
error = g_access(ump->um_cp, 0, -1, 0);
g_topology_unlock();
PICKUP_GIANT();
ump->um_fsckpid = 0;
}
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
if (fs->fs_ronly == 0 &&
vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0)) {
/*
* Flush any dirty data and suspend filesystem.
*/
if ((error = vn_start_write(NULL, &mp, V_WAIT)) != 0)
return (error);
for (;;) {
vn_finished_write(mp);
if ((error = vfs_write_suspend(mp)) != 0)
return (error);
MNT_ILOCK(mp);
if (mp->mnt_kern_flag & MNTK_SUSPENDED) {
/*
* Allow the secondary writes
* to proceed.
*/
mp->mnt_kern_flag &= ~(MNTK_SUSPENDED |
MNTK_SUSPEND2);
wakeup(&mp->mnt_flag);
MNT_IUNLOCK(mp);
/*
* Allow the curthread to
* ignore the suspension to
* synchronize on-disk state.
*/
td->td_pflags |= TDP_IGNSUSP;
break;
}
MNT_IUNLOCK(mp);
vn_start_write(NULL, &mp, V_WAIT);
}
/*
* Check for and optionally get rid of files open
* for writing.
*/
1994-05-24 10:09:53 +00:00
flags = WRITECLOSE;
if (mp->mnt_flag & MNT_FORCE)
flags |= FORCECLOSE;
if (MOUNTEDSOFTDEP(mp)) {
error = softdep_flushfiles(mp, flags, td);
} else {
error = ffs_flushfiles(mp, flags, td);
}
if (error) {
vfs_write_resume(mp);
return (error);
}
if (fs->fs_pendingblocks != 0 ||
fs->fs_pendinginodes != 0) {
printf("WARNING: %s Update error: blocks %jd "
"files %d\n", fs->fs_fsmnt,
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
(intmax_t)fs->fs_pendingblocks,
fs->fs_pendinginodes);
fs->fs_pendingblocks = 0;
fs->fs_pendinginodes = 0;
}
if ((fs->fs_flags & (FS_UNCLEAN | FS_NEEDSFSCK)) == 0)
fs->fs_clean = 1;
if ((error = ffs_sbupdate(ump, MNT_WAIT, 0)) != 0) {
fs->fs_ronly = 0;
fs->fs_clean = 0;
vfs_write_resume(mp);
return (error);
}
if (MOUNTEDSOFTDEP(mp))
softdep_unmount(mp);
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
DROP_GIANT();
g_topology_lock();
/*
* Drop our write and exclusive access.
*/
g_access(ump->um_cp, 0, -1, -1);
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
g_topology_unlock();
PICKUP_GIANT();
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
fs->fs_ronly = 1;
MNT_ILOCK(mp);
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
mp->mnt_flag |= MNT_RDONLY;
MNT_IUNLOCK(mp);
/*
* Allow the writers to note that filesystem
* is ro now.
*/
vfs_write_resume(mp);
}
if ((mp->mnt_flag & MNT_RELOAD) &&
(error = ffs_reload(mp, td)) != 0)
return (error);
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
if (fs->fs_ronly &&
!vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0)) {
/*
* If we are running a checker, do not allow upgrade.
*/
if (ump->um_fsckpid > 0) {
vfs_mount_error(mp,
"Active checker, cannot upgrade to write");
return (EINVAL);
}
/*
* If upgrade to read-write by non-root, then verify
* that user has necessary permissions on the device.
*/
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
error = VOP_ACCESS(devvp, VREAD | VWRITE,
td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
VOP_UNLOCK(devvp, 0);
return (error);
}
VOP_UNLOCK(devvp, 0);
fs->fs_flags &= ~FS_UNCLEAN;
if (fs->fs_clean == 0) {
fs->fs_flags |= FS_UNCLEAN;
if ((mp->mnt_flag & MNT_FORCE) ||
((fs->fs_flags &
(FS_SUJ | FS_NEEDSFSCK)) == 0 &&
(fs->fs_flags & FS_DOSOFTDEP))) {
printf("WARNING: %s was not properly "
"dismounted\n", fs->fs_fsmnt);
} else {
vfs_mount_error(mp,
"R/W mount of %s denied. %s.%s",
fs->fs_fsmnt,
"Filesystem is not clean - run fsck",
(fs->fs_flags & FS_SUJ) == 0 ? "" :
" Forced mount will invalidate"
" journal contents");
return (EPERM);
}
}
DROP_GIANT();
g_topology_lock();
/*
* Request exclusive write access.
*/
error = g_access(ump->um_cp, 0, 1, 1);
g_topology_unlock();
PICKUP_GIANT();
if (error)
return (error);
if ((error = vn_start_write(NULL, &mp, V_WAIT)) != 0)
return (error);
fs->fs_ronly = 0;
MNT_ILOCK(mp);
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
mp->mnt_flag &= ~MNT_RDONLY;
MNT_IUNLOCK(mp);
fs->fs_mtime = time_second;
/* check to see if we need to start softdep */
if ((fs->fs_flags & FS_DOSOFTDEP) &&
(error = softdep_mount(devvp, mp, fs, td->td_ucred))){
vn_finished_write(mp);
return (error);
}
fs->fs_clean = 0;
if ((error = ffs_sbupdate(ump, MNT_WAIT, 0)) != 0) {
vn_finished_write(mp);
return (error);
}
if (fs->fs_snapinum[0] != 0)
ffs_snapshot_mount(mp);
vn_finished_write(mp);
}
/*
* Soft updates is incompatible with "async",
* so if we are doing softupdates stop the user
* from setting the async flag in an update.
2010-09-17 09:14:40 +00:00
* Softdep_mount() clears it in an initial mount
* or ro->rw remount.
*/
if (MOUNTEDSOFTDEP(mp)) {
/* XXX: Reset too late ? */
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_ASYNC;
MNT_IUNLOCK(mp);
}
/*
* Keep MNT_ACLS flag if it is stored in superblock.
*/
if ((fs->fs_flags & FS_ACLS) != 0) {
/* XXX: Set too late ? */
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_ACLS;
MNT_IUNLOCK(mp);
}
if ((fs->fs_flags & FS_NFS4ACLS) != 0) {
/* XXX: Set too late ? */
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_NFS4ACLS;
MNT_IUNLOCK(mp);
}
/*
* If this is a request from fsck to clean up the filesystem,
* then allow the specified pid to proceed.
*/
if (fsckpid > 0) {
if (ump->um_fsckpid != 0) {
vfs_mount_error(mp,
"Active checker already running on %s",
fs->fs_fsmnt);
return (EINVAL);
}
KASSERT(MOUNTEDSOFTDEP(mp) == 0,
("soft updates enabled on read-only file system"));
DROP_GIANT();
g_topology_lock();
/*
* Request write access.
*/
error = g_access(ump->um_cp, 0, 1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error) {
vfs_mount_error(mp,
"Checker activation failed on %s",
fs->fs_fsmnt);
return (error);
}
ump->um_fsckpid = fsckpid;
if (fs->fs_snapinum[0] != 0)
ffs_snapshot_mount(mp);
fs->fs_mtime = time_second;
fs->fs_fmod = 1;
fs->fs_clean = 0;
(void) ffs_sbupdate(ump, MNT_WAIT, 0);
}
2010-09-17 09:14:40 +00:00
/*
* If this is a snapshot request, take the snapshot.
*/
if (mp->mnt_flag & MNT_SNAPSHOT)
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
return (ffs_snapshot(mp, fspec));
1994-05-24 10:09:53 +00:00
}
1994-05-24 10:09:53 +00:00
/*
* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible disk device.
1994-05-24 10:09:53 +00:00
*/
NDINIT(&ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspec, td);
if ((error = namei(&ndp)) != 0)
return (error);
NDFREE(&ndp, NDF_ONLY_PNBUF);
devvp = ndp.ni_vp;
if (!vn_isdisk(devvp, &error)) {
vput(devvp);
return (error);
}
/*
* If mount by non-root, then verify that user has necessary
* permissions on the device.
*/
accmode = VREAD;
if ((mp->mnt_flag & MNT_RDONLY) == 0)
accmode |= VWRITE;
error = VOP_ACCESS(devvp, accmode, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if (mp->mnt_flag & MNT_UPDATE) {
/*
* Update only
*
* If it's not the same vnode, or at least the same device
* then it's not correct.
*/
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
if (devvp->v_rdev != ump->um_devvp->v_rdev)
error = EINVAL; /* needs translation */
vput(devvp);
if (error)
return (error);
} else {
/*
* New mount
*
* We need the name for the mount point (also used for
* "last mounted on") copied in. If an error occurs,
* the mount point is discarded by the upper level code.
* Note that vfs_mount() populates f_mntonname for us.
*/
if ((error = ffs_mountfs(devvp, mp, td)) != 0) {
vrele(devvp);
return (error);
}
if (fsckpid > 0) {
KASSERT(MOUNTEDSOFTDEP(mp) == 0,
("soft updates enabled on read-only file system"));
ump = VFSTOUFS(mp);
fs = ump->um_fs;
DROP_GIANT();
g_topology_lock();
/*
* Request write access.
*/
error = g_access(ump->um_cp, 0, 1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error) {
printf("WARNING: %s: Checker activation "
"failed\n", fs->fs_fsmnt);
} else {
ump->um_fsckpid = fsckpid;
if (fs->fs_snapinum[0] != 0)
ffs_snapshot_mount(mp);
fs->fs_mtime = time_second;
fs->fs_clean = 0;
(void) ffs_sbupdate(ump, MNT_WAIT, 0);
}
}
1994-05-24 10:09:53 +00:00
}
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
vfs_mountedfrom(mp, fspec);
return (0);
1994-05-24 10:09:53 +00:00
}
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
/*
* Compatibility with old mount system call.
*/
static int
ffs_cmount(struct mntarg *ma, void *data, uint64_t flags)
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
{
struct ufs_args args;
struct export_args exp;
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
int error;
if (data == NULL)
return (EINVAL);
error = copyin(data, &args, sizeof args);
if (error)
return (error);
vfs_oexport_conv(&args.export, &exp);
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
ma = mount_argsu(ma, "from", args.fspec, MAXPATHLEN);
ma = mount_arg(ma, "export", &exp, sizeof(exp));
The remaining part of nmount/omount/rootfs mount changes. I cannot sensibly split the conversion of the remaining three filesystems out from the root mounting changes, so in one go: cd9660: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() nfs(client): Convert to nmount (the simple way, mount_nfs(8) is still necessary). Add omount compat shims. Drop COMPAT_PRELITE2 mount arg compatibility. ffs: Convert to nmount. Add omount compat shims. Remove dedicated rootfs mounting code. Use vfs_mountedfrom() Rely on vfs_mount.c calling VFS_STATFS() Remove vfs_omount() method, all filesystems are now converted. Remove MNTK_WANTRDWR, handling RO/RW conversions is a filesystem task, and they all do it now. Change rootmounting to use DEVFS trampoline: vfs_mount.c: Mount devfs on /. Devfs needs no 'from' so this is clean. symlink /dev to /. This makes it possible to lookup /dev/foo. Mount "real" root filesystem on /. Surgically move the devfs mountpoint from under the real root filesystem onto /dev in the real root filesystem. Remove now unnecessary getdiskbyname(). kern_init.c: Don't do devfs mounting and rootvnode assignment here, it was already handled by vfs_mount.c. Remove now unused bdevvp(), addaliasu() and addalias(). Put the few necessary lines in devfs where they belong. This eliminates the second-last source of bogo vnodes, leaving only the lemming-syncer. Remove rootdev variable, it doesn't give meaning in a global context and was not trustworth anyway. Correct information is provided by statfs(/).
2004-12-07 08:15:41 +00:00
error = kernel_mount(ma, flags);
return (error);
}
1994-05-24 10:09:53 +00:00
/*
* Reload all incore data for a filesystem (used after running fsck on
* the root filesystem and finding things to fix). The filesystem must
* be mounted read-only.
*
* Things to do to update the mount:
* 1) invalidate all cached meta-data.
* 2) re-read superblock from disk.
* 3) re-read summary information from disk.
* 4) invalidate all inactive vnodes.
* 5) invalidate all cached file data.
* 6) re-read inode data for all active vnodes.
*/
static int
ffs_reload(struct mount *mp, struct thread *td)
1994-05-24 10:09:53 +00:00
{
struct vnode *vp, *mvp, *devvp;
1994-05-24 10:09:53 +00:00
struct inode *ip;
void *space;
1994-05-24 10:09:53 +00:00
struct buf *bp;
struct fs *fs, *newfs;
struct ufsmount *ump;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
ufs2_daddr_t sblockloc;
1994-05-24 10:09:53 +00:00
int i, blks, size, error;
int32_t *lp;
1994-05-24 10:09:53 +00:00
if ((mp->mnt_flag & MNT_RDONLY) == 0)
1994-05-24 10:09:53 +00:00
return (EINVAL);
ump = VFSTOUFS(mp);
1994-05-24 10:09:53 +00:00
/*
* Step 1: invalidate all cached meta-data.
*/
devvp = VFSTOUFS(mp)->um_devvp;
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
if (vinvalbuf(devvp, 0, 0, 0) != 0)
1994-05-24 10:09:53 +00:00
panic("ffs_reload: dirty1");
VOP_UNLOCK(devvp, 0);
1994-05-24 10:09:53 +00:00
/*
* Step 2: re-read superblock from disk.
*/
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs = VFSTOUFS(mp)->um_fs;
if ((error = bread(devvp, btodb(fs->fs_sblockloc), fs->fs_sbsize,
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
NOCRED, &bp)) != 0)
1994-05-24 10:09:53 +00:00
return (error);
newfs = (struct fs *)bp->b_data;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
if ((newfs->fs_magic != FS_UFS1_MAGIC &&
newfs->fs_magic != FS_UFS2_MAGIC) ||
newfs->fs_bsize > MAXBSIZE ||
newfs->fs_bsize < sizeof(struct fs)) {
brelse(bp);
return (EIO); /* XXX needs translation */
1994-05-24 10:09:53 +00:00
}
/*
* Copy pointer fields back into superblock before copying in XXX
* new superblock. These should really be in the ufsmount. XXX
* Note that important parameters (eg fs_ncg) are unchanged.
*/
newfs->fs_csp = fs->fs_csp;
newfs->fs_maxcluster = fs->fs_maxcluster;
newfs->fs_contigdirs = fs->fs_contigdirs;
newfs->fs_active = fs->fs_active;
/* The file system is still read-only. */
newfs->fs_ronly = 1;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
sblockloc = fs->fs_sblockloc;
bcopy(newfs, fs, (u_int)fs->fs_sbsize);
1994-05-24 10:09:53 +00:00
brelse(bp);
mp->mnt_maxsymlinklen = fs->fs_maxsymlinklen;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
ffs_oldfscompat_read(fs, VFSTOUFS(mp), sblockloc);
UFS_LOCK(ump);
if (fs->fs_pendingblocks != 0 || fs->fs_pendinginodes != 0) {
printf("WARNING: %s: reload pending error: blocks %jd "
"files %d\n", fs->fs_fsmnt, (intmax_t)fs->fs_pendingblocks,
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_pendinginodes);
fs->fs_pendingblocks = 0;
fs->fs_pendinginodes = 0;
}
UFS_UNLOCK(ump);
1994-05-24 10:09:53 +00:00
/*
* Step 3: re-read summary information from disk.
*/
size = fs->fs_cssize;
blks = howmany(size, fs->fs_fsize);
if (fs->fs_contigsumsize > 0)
size += fs->fs_ncg * sizeof(int32_t);
size += fs->fs_ncg * sizeof(u_int8_t);
free(fs->fs_csp, M_UFSMNT);
space = malloc((u_long)size, M_UFSMNT, M_WAITOK);
fs->fs_csp = space;
1994-05-24 10:09:53 +00:00
for (i = 0; i < blks; i += fs->fs_frag) {
size = fs->fs_bsize;
if (i + fs->fs_frag > blks)
size = (blks - i) * fs->fs_fsize;
error = bread(devvp, fsbtodb(fs, fs->fs_csaddr + i), size,
NOCRED, &bp);
if (error)
1994-05-24 10:09:53 +00:00
return (error);
bcopy(bp->b_data, space, (u_int)size);
space = (char *)space + size;
1994-05-24 10:09:53 +00:00
brelse(bp);
}
/*
* We no longer know anything about clusters per cylinder group.
*/
if (fs->fs_contigsumsize > 0) {
fs->fs_maxcluster = lp = space;
for (i = 0; i < fs->fs_ncg; i++)
*lp++ = fs->fs_contigsumsize;
space = lp;
}
size = fs->fs_ncg * sizeof(u_int8_t);
fs->fs_contigdirs = (u_int8_t *)space;
bzero(fs->fs_contigdirs, size);
1994-05-24 10:09:53 +00:00
loop:
MNT_VNODE_FOREACH_ALL(vp, mp, mvp) {
1994-05-24 10:09:53 +00:00
/*
* Step 4: invalidate all cached file data.
1994-05-24 10:09:53 +00:00
*/
if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) {
MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
1994-05-24 10:09:53 +00:00
goto loop;
}
if (vinvalbuf(vp, 0, 0, 0))
1994-05-24 10:09:53 +00:00
panic("ffs_reload: dirty2");
/*
* Step 5: re-read inode data for all active vnodes.
1994-05-24 10:09:53 +00:00
*/
ip = VTOI(vp);
error =
1994-05-24 10:09:53 +00:00
bread(devvp, fsbtodb(fs, ino_to_fsba(fs, ip->i_number)),
(int)fs->fs_bsize, NOCRED, &bp);
if (error) {
VOP_UNLOCK(vp, 0);
vrele(vp);
MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
1994-05-24 10:09:53 +00:00
return (error);
}
ffs_load_inode(bp, ip, fs, ip->i_number);
ip->i_effnlink = ip->i_nlink;
1994-05-24 10:09:53 +00:00
brelse(bp);
VOP_UNLOCK(vp, 0);
vrele(vp);
1994-05-24 10:09:53 +00:00
}
return (0);
}
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
/*
* Possible superblock locations ordered from most to least likely.
*/
static int sblock_try[] = SBLOCKSEARCH;
1994-05-24 10:09:53 +00:00
/*
* Common code for mount and mountroot
*/
static int
ffs_mountfs(devvp, mp, td)
struct vnode *devvp;
1994-05-24 10:09:53 +00:00
struct mount *mp;
struct thread *td;
1994-05-24 10:09:53 +00:00
{
struct ufsmount *ump;
1994-05-24 10:09:53 +00:00
struct buf *bp;
struct fs *fs;
struct cdev *dev;
void *space;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
ufs2_daddr_t sblockloc;
int error, i, blks, size, ronly;
int32_t *lp;
struct ucred *cred;
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
struct g_consumer *cp;
struct mount *nmp;
1994-05-24 10:09:53 +00:00
bp = NULL;
ump = NULL;
cred = td ? td->td_ucred : NOCRED;
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
ronly = (mp->mnt_flag & MNT_RDONLY) != 0;
dev = devvp->v_rdev;
dev_ref(dev);
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
DROP_GIANT();
g_topology_lock();
error = g_vfs_open(devvp, &cp, "ffs", ronly ? 0 : 1);
g_topology_unlock();
PICKUP_GIANT();
VOP_UNLOCK(devvp, 0);
if (error)
goto out;
if (devvp->v_rdev->si_iosize_max != 0)
mp->mnt_iosize_max = devvp->v_rdev->si_iosize_max;
if (mp->mnt_iosize_max > MAXPHYS)
mp->mnt_iosize_max = MAXPHYS;
devvp->v_bufobj.bo_ops = &ffs_ops;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs = NULL;
sblockloc = 0;
/*
* Try reading the superblock in each of its possible locations.
*/
for (i = 0; sblock_try[i] != -1; i++) {
if ((SBLOCKSIZE % cp->provider->sectorsize) != 0) {
error = EINVAL;
vfs_mount_error(mp,
"Invalid sectorsize %d for superblock size %d",
cp->provider->sectorsize, SBLOCKSIZE);
goto out;
}
if ((error = bread(devvp, btodb(sblock_try[i]), SBLOCKSIZE,
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
cred, &bp)) != 0)
goto out;
fs = (struct fs *)bp->b_data;
sblockloc = sblock_try[i];
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
if ((fs->fs_magic == FS_UFS1_MAGIC ||
(fs->fs_magic == FS_UFS2_MAGIC &&
(fs->fs_sblockloc == sblockloc ||
(fs->fs_old_flags & FS_FLAGS_UPDATED) == 0))) &&
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_bsize <= MAXBSIZE &&
fs->fs_bsize >= sizeof(struct fs))
break;
brelse(bp);
bp = NULL;
}
if (sblock_try[i] == -1) {
1994-05-24 10:09:53 +00:00
error = EINVAL; /* XXX needs translation */
goto out;
}
fs->fs_fmod = 0;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_flags &= ~FS_INDEXDIRS; /* no support for directory indicies */
fs->fs_flags &= ~FS_UNCLEAN;
if (fs->fs_clean == 0) {
fs->fs_flags |= FS_UNCLEAN;
if (ronly || (mp->mnt_flag & MNT_FORCE) ||
((fs->fs_flags & (FS_SUJ | FS_NEEDSFSCK)) == 0 &&
(fs->fs_flags & FS_DOSOFTDEP))) {
2010-09-17 09:14:40 +00:00
printf("WARNING: %s was not properly dismounted\n",
fs->fs_fsmnt);
} else {
vfs_mount_error(mp, "R/W mount of %s denied. %s%s",
fs->fs_fsmnt, "Filesystem is not clean - run fsck.",
(fs->fs_flags & FS_SUJ) == 0 ? "" :
" Forced mount will invalidate journal contents");
error = EPERM;
goto out;
}
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
if ((fs->fs_pendingblocks != 0 || fs->fs_pendinginodes != 0) &&
(mp->mnt_flag & MNT_FORCE)) {
printf("WARNING: %s: lost blocks %jd files %d\n",
fs->fs_fsmnt, (intmax_t)fs->fs_pendingblocks,
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_pendinginodes);
fs->fs_pendingblocks = 0;
fs->fs_pendinginodes = 0;
}
}
if (fs->fs_pendingblocks != 0 || fs->fs_pendinginodes != 0) {
printf("WARNING: %s: mount pending error: blocks %jd "
"files %d\n", fs->fs_fsmnt, (intmax_t)fs->fs_pendingblocks,
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_pendinginodes);
fs->fs_pendingblocks = 0;
fs->fs_pendinginodes = 0;
}
if ((fs->fs_flags & FS_GJOURNAL) != 0) {
#ifdef UFS_GJOURNAL
/*
* Get journal provider name.
*/
size = 1024;
mp->mnt_gjprovider = malloc(size, M_UFSMNT, M_WAITOK);
if (g_io_getattr("GJOURNAL::provider", cp, &size,
mp->mnt_gjprovider) == 0) {
mp->mnt_gjprovider = realloc(mp->mnt_gjprovider, size,
M_UFSMNT, M_WAITOK);
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_GJOURNAL;
MNT_IUNLOCK(mp);
} else {
printf("WARNING: %s: GJOURNAL flag on fs "
"but no gjournal provider below\n",
mp->mnt_stat.f_mntonname);
free(mp->mnt_gjprovider, M_UFSMNT);
mp->mnt_gjprovider = NULL;
}
#else
printf("WARNING: %s: GJOURNAL flag on fs but no "
"UFS_GJOURNAL support\n", mp->mnt_stat.f_mntonname);
#endif
} else {
mp->mnt_gjprovider = NULL;
}
ump = malloc(sizeof *ump, M_UFSMNT, M_WAITOK | M_ZERO);
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
ump->um_cp = cp;
ump->um_bo = &devvp->v_bufobj;
ump->um_fs = malloc((u_long)fs->fs_sbsize, M_UFSMNT, M_WAITOK);
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
if (fs->fs_magic == FS_UFS1_MAGIC) {
ump->um_fstype = UFS1;
ump->um_balloc = ffs_balloc_ufs1;
} else {
ump->um_fstype = UFS2;
ump->um_balloc = ffs_balloc_ufs2;
}
ump->um_blkatoff = ffs_blkatoff;
ump->um_truncate = ffs_truncate;
ump->um_update = ffs_update;
ump->um_valloc = ffs_valloc;
ump->um_vfree = ffs_vfree;
ump->um_ifree = ffs_ifree;
ump->um_rdonly = ffs_rdonly;
ump->um_snapgone = ffs_snapgone;
mtx_init(UFS_MTX(ump), "FFS", "FFS Lock", MTX_DEF);
1994-05-24 10:09:53 +00:00
bcopy(bp->b_data, ump->um_fs, (u_int)fs->fs_sbsize);
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
if (fs->fs_sbsize < SBLOCKSIZE)
bp->b_flags |= B_INVAL | B_NOCACHE;
1994-05-24 10:09:53 +00:00
brelse(bp);
bp = NULL;
fs = ump->um_fs;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
ffs_oldfscompat_read(fs, ump, sblockloc);
1994-05-24 10:09:53 +00:00
fs->fs_ronly = ronly;
size = fs->fs_cssize;
blks = howmany(size, fs->fs_fsize);
if (fs->fs_contigsumsize > 0)
size += fs->fs_ncg * sizeof(int32_t);
Directory layout preference improvements from Grigoriy Orlov <gluk@ptci.ru>. His description of the problem and solution follow. My own tests show speedups on typical filesystem intensive workloads of 5% to 12% which is very impressive considering the small amount of code change involved. ------ One day I noticed that some file operations run much faster on small file systems then on big ones. I've looked at the ffs algorithms, thought about them, and redesigned the dirpref algorithm. First I want to describe the results of my tests. These results are old and I have improved the algorithm after these tests were done. Nevertheless they show how big the perfomance speedup may be. I have done two file/directory intensive tests on a two OpenBSD systems with old and new dirpref algorithm. The first test is "tar -xzf ports.tar.gz", the second is "rm -rf ports". The ports.tar.gz file is the ports collection from the OpenBSD 2.8 release. It contains 6596 directories and 13868 files. The test systems are: 1. Celeron-450, 128Mb, two IDE drives, the system at wd0, file system for test is at wd1. Size of test file system is 8 Gb, number of cg=991, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=35 2. PIII-600, 128Mb, two IBM DTLA-307045 IDE drives at i815e, the system at wd0, file system for test is at wd1. Size of test file system is 40 Gb, number of cg=5324, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=50 You can get more info about the test systems and methods at: http://www.ptci.ru/gluk/dirpref/old/dirpref.html Test Results tar -xzf ports.tar.gz rm -rf ports mode old dirpref new dirpref speedup old dirprefnew dirpref speedup First system normal 667 472 1.41 477 331 1.44 async 285 144 1.98 130 14 9.29 sync 768 616 1.25 477 334 1.43 softdep 413 252 1.64 241 38 6.34 Second system normal 329 81 4.06 263.5 93.5 2.81 async 302 25.7 11.75 112 2.26 49.56 sync 281 57.0 4.93 263 90.5 2.9 softdep 341 40.6 8.4 284 4.76 59.66 "old dirpref" and "new dirpref" columns give a test time in seconds. speedup - speed increasement in times, ie. old dirpref / new dirpref. ------ Algorithm description The old dirpref algorithm is described in comments: /* * Find a cylinder to place a directory. * * The policy implemented by this algorithm is to select from * among those cylinder groups with above the average number of * free inodes, the one with the smallest number of directories. */ A new directory is allocated in a different cylinder groups than its parent directory resulting in a directory tree that is spreaded across all the cylinder groups. This spreading out results in a non-optimal access to the directories and files. When we have a small filesystem it is not a problem but when the filesystem is big then perfomance degradation becomes very apparent. What I mean by a big file system ? 1. A big filesystem is a filesystem which occupy 20-30 or more percent of total drive space, i.e. first and last cylinder are physically located relatively far from each other. 2. It has a relatively large number of cylinder groups, for example more cylinder groups than 50% of the buffers in the buffer cache. The first results in long access times, while the second results in many buffers being used by metadata operations. Such operations use cylinder group blocks and on-disk inode blocks. The cylinder group block (fs->fs_cblkno) contains struct cg, inode and block bit maps. It is 2k in size for the default filesystem parameters. If new and parent directories are located in different cylinder groups then the system performs more input/output operations and uses more buffers. On filesystems with many cylinder groups, lots of cache buffers are used for metadata operations. My solution for this problem is very simple. I allocate many directories in one cylinder group. I also do some things, so that the new allocation method does not cause excessive fragmentation and all directory inodes will not be located at a location far from its file's inodes and data. The algorithm is: /* * Find a cylinder group to place a directory. * * The policy implemented by this algorithm is to allocate a * directory inode in the same cylinder group as its parent * directory, but also to reserve space for its files inodes * and data. Restrict the number of directories which may be * allocated one after another in the same cylinder group * without intervening allocation of files. * * If we allocate a first level directory then force allocation * in another cylinder group. */ My early versions of dirpref give me a good results for a wide range of file operations and different filesystem capacities except one case: those applications that create their entire directory structure first and only later fill this structure with files. My solution for such and similar cases is to limit a number of directories which may be created one after another in the same cylinder group without intervening file creations. For this purpose, I allocate an array of counters at mount time. This array is linked to the superblock fs->fs_contigdirs[cg]. Each time a directory is created the counter increases and each time a file is created the counter decreases. A 60Gb filesystem with 8mb/cg requires 10kb of memory for the counters array. The maxcontigdirs is a maximum number of directories which may be created without an intervening file creation. I found in my tests that the best performance occurs when I restrict the number of directories in one cylinder group such that all its files may be located in the same cylinder group. There may be some deterioration in performance if all the file inodes are in the same cylinder group as its containing directory, but their data partially resides in a different cylinder group. The maxcontigdirs value is calculated to try to prevent this condition. Since there is no way to know how many files and directories will be allocated later I added two optimization parameters in superblock/tunefs. They are: int32_t fs_avgfilesize; /* expected average file size */ int32_t fs_avgfpdir; /* expected # of files per directory */ These parameters have reasonable defaults but may be tweeked for special uses of a filesystem. They are only necessary in rare cases like better tuning a filesystem being used to store a squid cache. I have been using this algorithm for about 3 months. I have done a lot of testing on filesystems with different capacities, average filesize, average number of files per directory, and so on. I think this algorithm has no negative impact on filesystem perfomance. It works better than the default one in all cases. The new dirpref will greatly improve untarring/removing/coping of big directories, decrease load on cvs servers and much more. The new dirpref doesn't speedup a compilation process, but also doesn't slow it down. Obtained from: Grigoriy Orlov <gluk@ptci.ru>
2001-04-10 08:38:59 +00:00
size += fs->fs_ncg * sizeof(u_int8_t);
space = malloc((u_long)size, M_UFSMNT, M_WAITOK);
fs->fs_csp = space;
1994-05-24 10:09:53 +00:00
for (i = 0; i < blks; i += fs->fs_frag) {
size = fs->fs_bsize;
if (i + fs->fs_frag > blks)
size = (blks - i) * fs->fs_fsize;
if ((error = bread(devvp, fsbtodb(fs, fs->fs_csaddr + i), size,
cred, &bp)) != 0) {
free(fs->fs_csp, M_UFSMNT);
1994-05-24 10:09:53 +00:00
goto out;
}
bcopy(bp->b_data, space, (u_int)size);
space = (char *)space + size;
1994-05-24 10:09:53 +00:00
brelse(bp);
bp = NULL;
}
if (fs->fs_contigsumsize > 0) {
fs->fs_maxcluster = lp = space;
for (i = 0; i < fs->fs_ncg; i++)
*lp++ = fs->fs_contigsumsize;
space = lp;
}
Directory layout preference improvements from Grigoriy Orlov <gluk@ptci.ru>. His description of the problem and solution follow. My own tests show speedups on typical filesystem intensive workloads of 5% to 12% which is very impressive considering the small amount of code change involved. ------ One day I noticed that some file operations run much faster on small file systems then on big ones. I've looked at the ffs algorithms, thought about them, and redesigned the dirpref algorithm. First I want to describe the results of my tests. These results are old and I have improved the algorithm after these tests were done. Nevertheless they show how big the perfomance speedup may be. I have done two file/directory intensive tests on a two OpenBSD systems with old and new dirpref algorithm. The first test is "tar -xzf ports.tar.gz", the second is "rm -rf ports". The ports.tar.gz file is the ports collection from the OpenBSD 2.8 release. It contains 6596 directories and 13868 files. The test systems are: 1. Celeron-450, 128Mb, two IDE drives, the system at wd0, file system for test is at wd1. Size of test file system is 8 Gb, number of cg=991, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=35 2. PIII-600, 128Mb, two IBM DTLA-307045 IDE drives at i815e, the system at wd0, file system for test is at wd1. Size of test file system is 40 Gb, number of cg=5324, size of cg is 8m, block size = 8k, fragment size = 1k OpenBSD-current from Dec 2000 with BUFCACHEPERCENT=50 You can get more info about the test systems and methods at: http://www.ptci.ru/gluk/dirpref/old/dirpref.html Test Results tar -xzf ports.tar.gz rm -rf ports mode old dirpref new dirpref speedup old dirprefnew dirpref speedup First system normal 667 472 1.41 477 331 1.44 async 285 144 1.98 130 14 9.29 sync 768 616 1.25 477 334 1.43 softdep 413 252 1.64 241 38 6.34 Second system normal 329 81 4.06 263.5 93.5 2.81 async 302 25.7 11.75 112 2.26 49.56 sync 281 57.0 4.93 263 90.5 2.9 softdep 341 40.6 8.4 284 4.76 59.66 "old dirpref" and "new dirpref" columns give a test time in seconds. speedup - speed increasement in times, ie. old dirpref / new dirpref. ------ Algorithm description The old dirpref algorithm is described in comments: /* * Find a cylinder to place a directory. * * The policy implemented by this algorithm is to select from * among those cylinder groups with above the average number of * free inodes, the one with the smallest number of directories. */ A new directory is allocated in a different cylinder groups than its parent directory resulting in a directory tree that is spreaded across all the cylinder groups. This spreading out results in a non-optimal access to the directories and files. When we have a small filesystem it is not a problem but when the filesystem is big then perfomance degradation becomes very apparent. What I mean by a big file system ? 1. A big filesystem is a filesystem which occupy 20-30 or more percent of total drive space, i.e. first and last cylinder are physically located relatively far from each other. 2. It has a relatively large number of cylinder groups, for example more cylinder groups than 50% of the buffers in the buffer cache. The first results in long access times, while the second results in many buffers being used by metadata operations. Such operations use cylinder group blocks and on-disk inode blocks. The cylinder group block (fs->fs_cblkno) contains struct cg, inode and block bit maps. It is 2k in size for the default filesystem parameters. If new and parent directories are located in different cylinder groups then the system performs more input/output operations and uses more buffers. On filesystems with many cylinder groups, lots of cache buffers are used for metadata operations. My solution for this problem is very simple. I allocate many directories in one cylinder group. I also do some things, so that the new allocation method does not cause excessive fragmentation and all directory inodes will not be located at a location far from its file's inodes and data. The algorithm is: /* * Find a cylinder group to place a directory. * * The policy implemented by this algorithm is to allocate a * directory inode in the same cylinder group as its parent * directory, but also to reserve space for its files inodes * and data. Restrict the number of directories which may be * allocated one after another in the same cylinder group * without intervening allocation of files. * * If we allocate a first level directory then force allocation * in another cylinder group. */ My early versions of dirpref give me a good results for a wide range of file operations and different filesystem capacities except one case: those applications that create their entire directory structure first and only later fill this structure with files. My solution for such and similar cases is to limit a number of directories which may be created one after another in the same cylinder group without intervening file creations. For this purpose, I allocate an array of counters at mount time. This array is linked to the superblock fs->fs_contigdirs[cg]. Each time a directory is created the counter increases and each time a file is created the counter decreases. A 60Gb filesystem with 8mb/cg requires 10kb of memory for the counters array. The maxcontigdirs is a maximum number of directories which may be created without an intervening file creation. I found in my tests that the best performance occurs when I restrict the number of directories in one cylinder group such that all its files may be located in the same cylinder group. There may be some deterioration in performance if all the file inodes are in the same cylinder group as its containing directory, but their data partially resides in a different cylinder group. The maxcontigdirs value is calculated to try to prevent this condition. Since there is no way to know how many files and directories will be allocated later I added two optimization parameters in superblock/tunefs. They are: int32_t fs_avgfilesize; /* expected average file size */ int32_t fs_avgfpdir; /* expected # of files per directory */ These parameters have reasonable defaults but may be tweeked for special uses of a filesystem. They are only necessary in rare cases like better tuning a filesystem being used to store a squid cache. I have been using this algorithm for about 3 months. I have done a lot of testing on filesystems with different capacities, average filesize, average number of files per directory, and so on. I think this algorithm has no negative impact on filesystem perfomance. It works better than the default one in all cases. The new dirpref will greatly improve untarring/removing/coping of big directories, decrease load on cvs servers and much more. The new dirpref doesn't speedup a compilation process, but also doesn't slow it down. Obtained from: Grigoriy Orlov <gluk@ptci.ru>
2001-04-10 08:38:59 +00:00
size = fs->fs_ncg * sizeof(u_int8_t);
fs->fs_contigdirs = (u_int8_t *)space;
bzero(fs->fs_contigdirs, size);
fs->fs_active = NULL;
2007-10-16 10:54:55 +00:00
mp->mnt_data = ump;
mp->mnt_stat.f_fsid.val[0] = fs->fs_id[0];
mp->mnt_stat.f_fsid.val[1] = fs->fs_id[1];
nmp = NULL;
2010-09-17 09:14:40 +00:00
if (fs->fs_id[0] == 0 || fs->fs_id[1] == 0 ||
(nmp = vfs_getvfs(&mp->mnt_stat.f_fsid))) {
if (nmp)
vfs_rel(nmp);
vfs_getnewfsid(mp);
}
1994-05-24 10:09:53 +00:00
mp->mnt_maxsymlinklen = fs->fs_maxsymlinklen;
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_LOCAL;
MNT_IUNLOCK(mp);
if ((fs->fs_flags & FS_MULTILABEL) != 0) {
#ifdef MAC
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_MULTILABEL;
MNT_IUNLOCK(mp);
#else
printf("WARNING: %s: multilabel flag on fs but "
"no MAC support\n", mp->mnt_stat.f_mntonname);
#endif
}
if ((fs->fs_flags & FS_ACLS) != 0) {
#ifdef UFS_ACL
MNT_ILOCK(mp);
if (mp->mnt_flag & MNT_NFS4ACLS)
printf("WARNING: %s: ACLs flag on fs conflicts with "
"\"nfsv4acls\" mount option; option ignored\n",
mp->mnt_stat.f_mntonname);
mp->mnt_flag &= ~MNT_NFS4ACLS;
mp->mnt_flag |= MNT_ACLS;
MNT_IUNLOCK(mp);
#else
2010-09-17 09:14:40 +00:00
printf("WARNING: %s: ACLs flag on fs but no ACLs support\n",
mp->mnt_stat.f_mntonname);
#endif
}
if ((fs->fs_flags & FS_NFS4ACLS) != 0) {
#ifdef UFS_ACL
MNT_ILOCK(mp);
if (mp->mnt_flag & MNT_ACLS)
printf("WARNING: %s: NFSv4 ACLs flag on fs conflicts "
"with \"acls\" mount option; option ignored\n",
mp->mnt_stat.f_mntonname);
mp->mnt_flag &= ~MNT_ACLS;
mp->mnt_flag |= MNT_NFS4ACLS;
MNT_IUNLOCK(mp);
#else
printf("WARNING: %s: NFSv4 ACLs flag on fs but no "
"ACLs support\n", mp->mnt_stat.f_mntonname);
#endif
}
if ((fs->fs_flags & FS_TRIM) != 0) {
size = sizeof(int);
if (g_io_getattr("GEOM::candelete", cp, &size,
&ump->um_candelete) == 0) {
if (!ump->um_candelete)
printf("WARNING: %s: TRIM flag on fs but disk "
"does not support TRIM\n",
mp->mnt_stat.f_mntonname);
} else {
printf("WARNING: %s: TRIM flag on fs but disk does "
"not confirm that it supports TRIM\n",
mp->mnt_stat.f_mntonname);
ump->um_candelete = 0;
}
}
1994-05-24 10:09:53 +00:00
ump->um_mountp = mp;
ump->um_dev = dev;
ump->um_devvp = devvp;
ump->um_nindir = fs->fs_nindir;
ump->um_bptrtodb = fs->fs_fsbtodb;
ump->um_seqinc = fs->fs_frag;
for (i = 0; i < MAXQUOTAS; i++)
ump->um_quotas[i] = NULLVP;
#ifdef UFS_EXTATTR
ufs_extattr_uepm_init(&ump->um_extattr);
#endif
/*
* Set FS local "last mounted on" information (NULL pad)
*/
bzero(fs->fs_fsmnt, MAXMNTLEN);
strlcpy(fs->fs_fsmnt, mp->mnt_stat.f_mntonname, MAXMNTLEN);
mp->mnt_stat.f_iosize = fs->fs_bsize;
if (mp->mnt_flag & MNT_ROOTFS) {
/*
* Root mount; update timestamp in mount structure.
* this will be used by the common root mount code
* to update the system clock.
*/
mp->mnt_time = fs->fs_time;
}
if (ronly == 0) {
fs->fs_mtime = time_second;
if ((fs->fs_flags & FS_DOSOFTDEP) &&
(error = softdep_mount(devvp, mp, fs, cred)) != 0) {
free(fs->fs_csp, M_UFSMNT);
ffs_flushfiles(mp, FORCECLOSE, td);
goto out;
}
if (devvp->v_type == VCHR && devvp->v_rdev != NULL)
devvp->v_rdev->si_mountpt = mp;
if (fs->fs_snapinum[0] != 0)
ffs_snapshot_mount(mp);
fs->fs_fmod = 1;
fs->fs_clean = 0;
(void) ffs_sbupdate(ump, MNT_WAIT, 0);
}
/*
* Initialize filesystem stat information in mount struct.
*/
MNT_ILOCK(mp);
mp->mnt_kern_flag |= MNTK_MPSAFE | MNTK_LOOKUP_SHARED |
MNTK_EXTENDED_SHARED;
MNT_IUNLOCK(mp);
#ifdef UFS_EXTATTR
#ifdef UFS_EXTATTR_AUTOSTART
/*
*
o Implement "options FFS_EXTATTR_AUTOSTART", which depends on "options FFS_EXTATTR". When extended attribute auto-starting is enabled, FFS will scan the .attribute directory off of the root of each file system, as it is mounted. If .attribute exists, EA support will be started for the file system. If there are files in the directory, FFS will attempt to start them as attribute backing files for attributes baring the same name. All attributes are started before access to the file system is permitted, so this permits race-free enabling of attributes. For attributes backing support for security features, such as ACLs, MAC, Capabilities, this is vital, as it prevents the file system attributes from getting out of sync as a result of file system operations between mount-time and the enabling of the extended attribute. The userland extattrctl tool will still function exactly as previously. Files must be placed directly in .attribute, which must be directly off of the file system root: symbolic links are not permitted. FFS_EXTATTR will continue to be able to function without FFS_EXTATTR_AUTOSTART for sites that do not want/require auto-starting. If you're using the UFS_ACL code available from www.TrustedBSD.org, using FFS_EXTATTR_AUTOSTART is recommended. o This support is implemented by adding an invocation of ufs_extattr_autostart() to ffs_mountfs(). In addition, several new supporting calls are introduced in ufs_extattr.c: ufs_extattr_autostart(): start EAs on the specified mount ufs_extattr_lookup(): given a directory and filename, return the vnode for the file. ufs_extattr_enable_with_open(): invoke ufs_extattr_enable() after doing the equililent of vn_open() on the passed file. ufs_extattr_iterate_directory(): iterate over a directory, invoking ufs_extattr_lookup() and ufs_extattr_enable_with_open() on each entry. o This feature is not widely tested, and therefore may contain bugs, caution is advised. Several changes are in the pipeline for this feature, including breaking out of EA namespaces into subdirectories of .attribute (this is waiting on the updated EA API), as well as a per-filesystem flag indicating whether or not EAs should be auto-started. This is required because administrators may not want .attribute auto-started on all file systems, especially if non-administrators have write access to the root of a file system. Obtained from: TrustedBSD Project
2001-03-14 05:32:31 +00:00
* Auto-starting does the following:
* - check for /.attribute in the fs, and extattr_start if so
* - for each file in .attribute, enable that file with
* an attribute of the same name.
* Not clear how to report errors -- probably eat them.
2002-05-16 21:28:32 +00:00
* This would all happen while the filesystem was busy/not
* available, so would effectively be "atomic".
*/
(void) ufs_extattr_autostart(mp, td);
#endif /* !UFS_EXTATTR_AUTOSTART */
#endif /* !UFS_EXTATTR */
1994-05-24 10:09:53 +00:00
return (0);
out:
if (bp)
brelse(bp);
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
if (cp != NULL) {
DROP_GIANT();
g_topology_lock();
g_vfs_close(cp);
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
g_topology_unlock();
PICKUP_GIANT();
}
1994-05-24 10:09:53 +00:00
if (ump) {
mtx_destroy(UFS_MTX(ump));
if (mp->mnt_gjprovider != NULL) {
free(mp->mnt_gjprovider, M_UFSMNT);
mp->mnt_gjprovider = NULL;
}
1994-05-24 10:09:53 +00:00
free(ump->um_fs, M_UFSMNT);
free(ump, M_UFSMNT);
2007-10-16 10:54:55 +00:00
mp->mnt_data = NULL;
1994-05-24 10:09:53 +00:00
}
dev_rel(dev);
1994-05-24 10:09:53 +00:00
return (error);
}
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
#include <sys/sysctl.h>
static int bigcgs = 0;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
SYSCTL_INT(_debug, OID_AUTO, bigcgs, CTLFLAG_RW, &bigcgs, 0, "");
1994-05-24 10:09:53 +00:00
/*
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
* Sanity checks for loading old filesystem superblocks.
* See ffs_oldfscompat_write below for unwound actions.
1994-05-24 10:09:53 +00:00
*
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
* XXX - Parts get retired eventually.
* Unfortunately new bits get added.
1994-05-24 10:09:53 +00:00
*/
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
static void
ffs_oldfscompat_read(fs, ump, sblockloc)
1994-05-24 10:09:53 +00:00
struct fs *fs;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
struct ufsmount *ump;
ufs2_daddr_t sblockloc;
1994-05-24 10:09:53 +00:00
{
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
off_t maxfilesize;
1994-05-24 10:09:53 +00:00
/*
* If not yet done, update fs_flags location and value of fs_sblockloc.
*/
if ((fs->fs_old_flags & FS_FLAGS_UPDATED) == 0) {
fs->fs_flags = fs->fs_old_flags;
fs->fs_old_flags |= FS_FLAGS_UPDATED;
fs->fs_sblockloc = sblockloc;
}
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
/*
* If not yet done, update UFS1 superblock with new wider fields.
*/
if (fs->fs_magic == FS_UFS1_MAGIC && fs->fs_maxbsize != fs->fs_bsize) {
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_maxbsize = fs->fs_bsize;
fs->fs_time = fs->fs_old_time;
fs->fs_size = fs->fs_old_size;
fs->fs_dsize = fs->fs_old_dsize;
fs->fs_csaddr = fs->fs_old_csaddr;
fs->fs_cstotal.cs_ndir = fs->fs_old_cstotal.cs_ndir;
fs->fs_cstotal.cs_nbfree = fs->fs_old_cstotal.cs_nbfree;
fs->fs_cstotal.cs_nifree = fs->fs_old_cstotal.cs_nifree;
fs->fs_cstotal.cs_nffree = fs->fs_old_cstotal.cs_nffree;
}
if (fs->fs_magic == FS_UFS1_MAGIC &&
fs->fs_old_inodefmt < FS_44INODEFMT) {
fs->fs_maxfilesize = ((uint64_t)1 << 31) - 1;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_qbmask = ~fs->fs_bmask;
fs->fs_qfmask = ~fs->fs_fmask;
}
if (fs->fs_magic == FS_UFS1_MAGIC) {
ump->um_savedmaxfilesize = fs->fs_maxfilesize;
maxfilesize = (uint64_t)0x80000000 * fs->fs_bsize - 1;
if (fs->fs_maxfilesize > maxfilesize)
fs->fs_maxfilesize = maxfilesize;
}
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
/* Compatibility for old filesystems */
if (fs->fs_avgfilesize <= 0)
fs->fs_avgfilesize = AVFILESIZ;
if (fs->fs_avgfpdir <= 0)
fs->fs_avgfpdir = AFPDIR;
if (bigcgs) {
fs->fs_save_cgsize = fs->fs_cgsize;
fs->fs_cgsize = fs->fs_bsize;
}
}
/*
* Unwinding superblock updates for old filesystems.
* See ffs_oldfscompat_read above for details.
*
* XXX - Parts get retired eventually.
* Unfortunately new bits get added.
*/
void
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
ffs_oldfscompat_write(fs, ump)
struct fs *fs;
struct ufsmount *ump;
{
/*
* Copy back UFS2 updated fields that UFS1 inspects.
*/
if (fs->fs_magic == FS_UFS1_MAGIC) {
fs->fs_old_time = fs->fs_time;
fs->fs_old_cstotal.cs_ndir = fs->fs_cstotal.cs_ndir;
fs->fs_old_cstotal.cs_nbfree = fs->fs_cstotal.cs_nbfree;
fs->fs_old_cstotal.cs_nifree = fs->fs_cstotal.cs_nifree;
fs->fs_old_cstotal.cs_nffree = fs->fs_cstotal.cs_nffree;
fs->fs_maxfilesize = ump->um_savedmaxfilesize;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
}
if (bigcgs) {
fs->fs_cgsize = fs->fs_save_cgsize;
fs->fs_save_cgsize = 0;
}
1994-05-24 10:09:53 +00:00
}
/*
* unmount system call
*/
static int
ffs_unmount(mp, mntflags)
1994-05-24 10:09:53 +00:00
struct mount *mp;
int mntflags;
{
struct thread *td;
struct ufsmount *ump = VFSTOUFS(mp);
struct fs *fs;
int error, flags, susp;
#ifdef UFS_EXTATTR
int e_restart;
#endif
1994-05-24 10:09:53 +00:00
flags = 0;
td = curthread;
fs = ump->um_fs;
susp = 0;
1994-05-24 10:09:53 +00:00
if (mntflags & MNT_FORCE) {
flags |= FORCECLOSE;
susp = fs->fs_ronly != 0;
}
#ifdef UFS_EXTATTR
if ((error = ufs_extattr_stop(mp, td))) {
if (error != EOPNOTSUPP)
printf("WARNING: unmount %s: ufs_extattr_stop "
"returned errno %d\n", mp->mnt_stat.f_mntonname,
error);
e_restart = 0;
} else {
ufs_extattr_uepm_destroy(&ump->um_extattr);
e_restart = 1;
}
#endif
if (susp) {
/*
* dounmount already called vn_start_write().
*/
for (;;) {
vn_finished_write(mp);
if ((error = vfs_write_suspend(mp)) != 0)
return (error);
MNT_ILOCK(mp);
if (mp->mnt_kern_flag & MNTK_SUSPENDED) {
mp->mnt_kern_flag &= ~(MNTK_SUSPENDED |
MNTK_SUSPEND2);
wakeup(&mp->mnt_flag);
MNT_IUNLOCK(mp);
td->td_pflags |= TDP_IGNSUSP;
break;
}
MNT_IUNLOCK(mp);
vn_start_write(NULL, &mp, V_WAIT);
}
}
if (MOUNTEDSOFTDEP(mp))
error = softdep_flushfiles(mp, flags, td);
else
error = ffs_flushfiles(mp, flags, td);
if (error != 0 && error != ENXIO)
goto fail;
UFS_LOCK(ump);
if (fs->fs_pendingblocks != 0 || fs->fs_pendinginodes != 0) {
printf("WARNING: unmount %s: pending error: blocks %jd "
"files %d\n", fs->fs_fsmnt, (intmax_t)fs->fs_pendingblocks,
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
fs->fs_pendinginodes);
fs->fs_pendingblocks = 0;
fs->fs_pendinginodes = 0;
}
UFS_UNLOCK(ump);
softdep_unmount(mp);
if (fs->fs_ronly == 0 || ump->um_fsckpid > 0) {
fs->fs_clean = fs->fs_flags & (FS_UNCLEAN|FS_NEEDSFSCK) ? 0 : 1;
error = ffs_sbupdate(ump, MNT_WAIT, 0);
if (error && error != ENXIO) {
fs->fs_clean = 0;
goto fail;
}
}
if (susp) {
vfs_write_resume(mp);
vn_start_write(NULL, &mp, V_WAIT);
}
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
DROP_GIANT();
g_topology_lock();
if (ump->um_fsckpid > 0) {
/*
* Return to normal read-only mode.
*/
error = g_access(ump->um_cp, 0, -1, 0);
ump->um_fsckpid = 0;
}
g_vfs_close(ump->um_cp);
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
g_topology_unlock();
PICKUP_GIANT();
if (ump->um_devvp->v_type == VCHR && ump->um_devvp->v_rdev != NULL)
ump->um_devvp->v_rdev->si_mountpt = NULL;
vrele(ump->um_devvp);
dev_rel(ump->um_dev);
mtx_destroy(UFS_MTX(ump));
if (mp->mnt_gjprovider != NULL) {
free(mp->mnt_gjprovider, M_UFSMNT);
mp->mnt_gjprovider = NULL;
}
free(fs->fs_csp, M_UFSMNT);
1994-05-24 10:09:53 +00:00
free(fs, M_UFSMNT);
free(ump, M_UFSMNT);
2007-10-16 10:54:55 +00:00
mp->mnt_data = NULL;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_LOCAL;
MNT_IUNLOCK(mp);
1994-05-24 10:09:53 +00:00
return (error);
fail:
if (susp) {
vfs_write_resume(mp);
vn_start_write(NULL, &mp, V_WAIT);
}
#ifdef UFS_EXTATTR
if (e_restart) {
ufs_extattr_uepm_init(&ump->um_extattr);
#ifdef UFS_EXTATTR_AUTOSTART
(void) ufs_extattr_autostart(mp, td);
#endif
}
#endif
return (error);
1994-05-24 10:09:53 +00:00
}
/*
* Flush out all the files in a filesystem.
*/
int
ffs_flushfiles(mp, flags, td)
struct mount *mp;
1994-05-24 10:09:53 +00:00
int flags;
struct thread *td;
1994-05-24 10:09:53 +00:00
{
struct ufsmount *ump;
int error;
1994-05-24 10:09:53 +00:00
ump = VFSTOUFS(mp);
#ifdef QUOTA
if (mp->mnt_flag & MNT_QUOTA) {
int i;
error = vflush(mp, 0, SKIPSYSTEM|flags, td);
if (error)
1994-05-24 10:09:53 +00:00
return (error);
for (i = 0; i < MAXQUOTAS; i++) {
quotaoff(td, mp, i);
1994-05-24 10:09:53 +00:00
}
/*
* Here we fall through to vflush again to ensure
* that we have gotten rid of all the system vnodes.
*/
}
#endif
ASSERT_VOP_LOCKED(ump->um_devvp, "ffs_flushfiles");
if (ump->um_devvp->v_vflag & VV_COPYONWRITE) {
if ((error = vflush(mp, 0, SKIPSYSTEM | flags, td)) != 0)
return (error);
ffs_snapshot_unmount(mp);
flags |= FORCECLOSE;
/*
* Here we fall through to vflush again to ensure
* that we have gotten rid of all the system vnodes.
*/
}
/*
* Flush all the files.
*/
if ((error = vflush(mp, 0, flags, td)) != 0)
return (error);
/*
* Flush filesystem metadata.
*/
vn_lock(ump->um_devvp, LK_EXCLUSIVE | LK_RETRY);
error = VOP_FSYNC(ump->um_devvp, MNT_WAIT, td);
VOP_UNLOCK(ump->um_devvp, 0);
1994-05-24 10:09:53 +00:00
return (error);
}
/*
2002-05-16 21:28:32 +00:00
* Get filesystem statistics.
1994-05-24 10:09:53 +00:00
*/
static int
ffs_statfs(mp, sbp)
1994-05-24 10:09:53 +00:00
struct mount *mp;
struct statfs *sbp;
1994-05-24 10:09:53 +00:00
{
struct ufsmount *ump;
struct fs *fs;
1994-05-24 10:09:53 +00:00
ump = VFSTOUFS(mp);
fs = ump->um_fs;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
if (fs->fs_magic != FS_UFS1_MAGIC && fs->fs_magic != FS_UFS2_MAGIC)
1994-05-24 10:09:53 +00:00
panic("ffs_statfs");
sbp->f_version = STATFS_VERSION;
1994-05-24 10:09:53 +00:00
sbp->f_bsize = fs->fs_fsize;
sbp->f_iosize = fs->fs_bsize;
sbp->f_blocks = fs->fs_dsize;
UFS_LOCK(ump);
1994-05-24 10:09:53 +00:00
sbp->f_bfree = fs->fs_cstotal.cs_nbfree * fs->fs_frag +
fs->fs_cstotal.cs_nffree + dbtofsb(fs, fs->fs_pendingblocks);
sbp->f_bavail = freespace(fs, fs->fs_minfree) +
dbtofsb(fs, fs->fs_pendingblocks);
1994-05-24 10:09:53 +00:00
sbp->f_files = fs->fs_ncg * fs->fs_ipg - ROOTINO;
sbp->f_ffree = fs->fs_cstotal.cs_nifree + fs->fs_pendinginodes;
UFS_UNLOCK(ump);
sbp->f_namemax = NAME_MAX;
1994-05-24 10:09:53 +00:00
return (0);
}
/*
* For a lazy sync, we only care about access times, quotas and the
* superblock. Other filesystem changes are already converted to
* cylinder group blocks or inode blocks updates and are written to
* disk by syncer.
*/
static int
ffs_sync_lazy(mp)
struct mount *mp;
{
struct vnode *mvp, *vp;
struct inode *ip;
struct thread *td;
int allerror, error;
allerror = 0;
td = curthread;
if ((mp->mnt_flag & MNT_NOATIME) != 0)
goto qupdate;
MNT_VNODE_FOREACH_ACTIVE(vp, mp, mvp) {
if (vp->v_type == VNON) {
VI_UNLOCK(vp);
continue;
}
ip = VTOI(vp);
/*
* The IN_ACCESS flag is converted to IN_MODIFIED by
* ufs_close() and ufs_getattr() by the calls to
* ufs_itimes_locked(), without subsequent UFS_UPDATE().
* Test also all the other timestamp flags too, to pick up
* any other cases that could be missed.
*/
if ((ip->i_flag & (IN_ACCESS | IN_CHANGE | IN_MODIFIED |
IN_UPDATE)) == 0) {
VI_UNLOCK(vp);
continue;
}
if ((error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK,
td)) != 0)
continue;
error = ffs_update(vp, 0);
if (error != 0)
allerror = error;
vput(vp);
}
qupdate:
#ifdef QUOTA
qsync(mp);
#endif
if (VFSTOUFS(mp)->um_fs->fs_fmod != 0 &&
(error = ffs_sbupdate(VFSTOUFS(mp), MNT_LAZY, 0)) != 0)
allerror = error;
return (allerror);
}
1994-05-24 10:09:53 +00:00
/*
* Go through the disk queues to initiate sandbagged IO;
* go through the inodes to write those that have been modified;
* initiate the writing of the super block if it has been modified.
*
* Note: we are always called with the filesystem marked busy using
* vfs_busy().
1994-05-24 10:09:53 +00:00
*/
static int
ffs_sync(mp, waitfor)
1994-05-24 10:09:53 +00:00
struct mount *mp;
int waitfor;
{
struct vnode *mvp, *vp, *devvp;
struct thread *td;
struct inode *ip;
struct ufsmount *ump = VFSTOUFS(mp);
struct fs *fs;
This patch corrects the first round of panics and hangs reported with the new snapshot code. Update addaliasu to correctly implement the semantics of the old checkalias function. When a device vnode first comes into existence, check to see if an anonymous vnode for the same device was created at boot time by bdevvp(). If so, adopt the bdevvp vnode rather than creating a new vnode for the device. This corrects a problem which caused the kernel to panic when taking a snapshot of the root filesystem. Change the calling convention of vn_write_suspend_wait() to be the same as vn_start_write(). Split out softdep_flushworklist() from softdep_flushfiles() so that it can be used to clear the work queue when suspending filesystem operations. Access to buffers becomes recursive so that snapshots can recursively traverse their indirect blocks using ffs_copyonwrite() when checking for the need for copy on write when flushing one of their own indirect blocks. This eliminates a deadlock between the syncer daemon and a process taking a snapshot. Ensure that softdep_process_worklist() can never block because of a snapshot being taken. This eliminates a problem with buffer starvation. Cleanup change in ffs_sync() which did not synchronously wait when MNT_WAIT was specified. The result was an unclean filesystem panic when doing forcible unmount with heavy filesystem I/O in progress. Return a zero'ed block when reading a block that was not in use at the time that a snapshot was taken. Normally, these blocks should never be read. However, the readahead code will occationally read them which can cause unexpected behavior. Clean up the debugging code that ensures that no blocks be written on a filesystem while it is suspended. Snapshots must explicitly label the blocks that they are writing during the suspension so that they do not cause a `write on suspended filesystem' panic. Reorganize ffs_copyonwrite() to eliminate a deadlock and also to prevent a race condition that would permit the same block to be copied twice. This change eliminates an unexpected soft updates inconsistency in fsck caused by the double allocation. Use bqrelse rather than brelse for buffers that will be needed soon again by the snapshot code. This improves snapshot performance.
2000-07-24 05:28:33 +00:00
int error, count, wait, lockreq, allerror = 0;
int suspend;
int suspended;
int secondary_writes;
int secondary_accwrites;
int softdep_deps;
int softdep_accdeps;
struct bufobj *bo;
1994-05-24 10:09:53 +00:00
In the original days of BSD, a sync was issued on every filesystem every 30 seconds. This spike in I/O caused the system to pause every 30 seconds which was quite annoying. So, the way that sync worked was changed so that when a vnode was first dirtied, it was put on a 30-second cleaning queue (see the syncer_workitem_pending queues in kern/vfs_subr.c). If the file has not been written or deleted after 30 seconds, the syncer pushes it out. As the syncer runs once per second, dirty files are trickled out slowly over the 30-second period instead of all at once by a call to sync(2). The one drawback to this is that it does not cover the filesystem metadata. To handle the metadata, vfs_allocate_syncvnode() is called to create a "filesystem syncer vnode" at mount time which cycles around the cleaning queue being sync'ed every 30 seconds. In the original design, the only things it would sync for UFS were the filesystem metadata: inode blocks, cylinder group bitmaps, and the superblock (e.g., by VOP_FSYNC'ing devvp, the device vnode from which the filesystem is mounted). Somewhere in its path to integration with FreeBSD the flushing of the filesystem syncer vnode got changed to sync every vnode associated with the filesystem. The result of this change is to return to the old filesystem-wide flush every 30-seconds behavior and makes the whole 30-second delay per vnode useless. This change goes back to the originally intended trickle out sync behavior. Key to ensuring that all the intended semantics are preserved (e.g., that all inode updates get flushed within a bounded period of time) is that all inode modifications get pushed to their corresponding inode blocks so that the metadata flush by the filesystem syncer vnode gets them to the disk in a timely way. Thanks to Konstantin Belousov (kib@) for doing the audit and commit -r231122 which ensures that all of these updates are being made. Reviewed by: kib Tested by: scottl MFC after: 2 weeks
2012-02-07 20:43:28 +00:00
wait = 0;
suspend = 0;
suspended = 0;
td = curthread;
1994-05-24 10:09:53 +00:00
fs = ump->um_fs;
if (fs->fs_fmod != 0 && fs->fs_ronly != 0 && ump->um_fsckpid == 0)
panic("%s: ffs_sync: modification on read-only filesystem",
fs->fs_fsmnt);
if (waitfor == MNT_LAZY)
return (ffs_sync_lazy(mp));
1994-05-24 10:09:53 +00:00
/*
* Write back each (modified) inode.
*/
lockreq = LK_EXCLUSIVE | LK_NOWAIT;
if (waitfor == MNT_SUSPEND) {
suspend = 1;
waitfor = MNT_WAIT;
}
This patch corrects the first round of panics and hangs reported with the new snapshot code. Update addaliasu to correctly implement the semantics of the old checkalias function. When a device vnode first comes into existence, check to see if an anonymous vnode for the same device was created at boot time by bdevvp(). If so, adopt the bdevvp vnode rather than creating a new vnode for the device. This corrects a problem which caused the kernel to panic when taking a snapshot of the root filesystem. Change the calling convention of vn_write_suspend_wait() to be the same as vn_start_write(). Split out softdep_flushworklist() from softdep_flushfiles() so that it can be used to clear the work queue when suspending filesystem operations. Access to buffers becomes recursive so that snapshots can recursively traverse their indirect blocks using ffs_copyonwrite() when checking for the need for copy on write when flushing one of their own indirect blocks. This eliminates a deadlock between the syncer daemon and a process taking a snapshot. Ensure that softdep_process_worklist() can never block because of a snapshot being taken. This eliminates a problem with buffer starvation. Cleanup change in ffs_sync() which did not synchronously wait when MNT_WAIT was specified. The result was an unclean filesystem panic when doing forcible unmount with heavy filesystem I/O in progress. Return a zero'ed block when reading a block that was not in use at the time that a snapshot was taken. Normally, these blocks should never be read. However, the readahead code will occationally read them which can cause unexpected behavior. Clean up the debugging code that ensures that no blocks be written on a filesystem while it is suspended. Snapshots must explicitly label the blocks that they are writing during the suspension so that they do not cause a `write on suspended filesystem' panic. Reorganize ffs_copyonwrite() to eliminate a deadlock and also to prevent a race condition that would permit the same block to be copied twice. This change eliminates an unexpected soft updates inconsistency in fsck caused by the double allocation. Use bqrelse rather than brelse for buffers that will be needed soon again by the snapshot code. This improves snapshot performance.
2000-07-24 05:28:33 +00:00
if (waitfor == MNT_WAIT) {
wait = 1;
lockreq = LK_EXCLUSIVE;
This patch corrects the first round of panics and hangs reported with the new snapshot code. Update addaliasu to correctly implement the semantics of the old checkalias function. When a device vnode first comes into existence, check to see if an anonymous vnode for the same device was created at boot time by bdevvp(). If so, adopt the bdevvp vnode rather than creating a new vnode for the device. This corrects a problem which caused the kernel to panic when taking a snapshot of the root filesystem. Change the calling convention of vn_write_suspend_wait() to be the same as vn_start_write(). Split out softdep_flushworklist() from softdep_flushfiles() so that it can be used to clear the work queue when suspending filesystem operations. Access to buffers becomes recursive so that snapshots can recursively traverse their indirect blocks using ffs_copyonwrite() when checking for the need for copy on write when flushing one of their own indirect blocks. This eliminates a deadlock between the syncer daemon and a process taking a snapshot. Ensure that softdep_process_worklist() can never block because of a snapshot being taken. This eliminates a problem with buffer starvation. Cleanup change in ffs_sync() which did not synchronously wait when MNT_WAIT was specified. The result was an unclean filesystem panic when doing forcible unmount with heavy filesystem I/O in progress. Return a zero'ed block when reading a block that was not in use at the time that a snapshot was taken. Normally, these blocks should never be read. However, the readahead code will occationally read them which can cause unexpected behavior. Clean up the debugging code that ensures that no blocks be written on a filesystem while it is suspended. Snapshots must explicitly label the blocks that they are writing during the suspension so that they do not cause a `write on suspended filesystem' panic. Reorganize ffs_copyonwrite() to eliminate a deadlock and also to prevent a race condition that would permit the same block to be copied twice. This change eliminates an unexpected soft updates inconsistency in fsck caused by the double allocation. Use bqrelse rather than brelse for buffers that will be needed soon again by the snapshot code. This improves snapshot performance.
2000-07-24 05:28:33 +00:00
}
lockreq |= LK_INTERLOCK | LK_SLEEPFAIL;
1994-05-24 10:09:53 +00:00
loop:
/* Grab snapshot of secondary write counts */
MNT_ILOCK(mp);
secondary_writes = mp->mnt_secondary_writes;
secondary_accwrites = mp->mnt_secondary_accwrites;
MNT_IUNLOCK(mp);
/* Grab snapshot of softdep dependency counts */
softdep_get_depcounts(mp, &softdep_deps, &softdep_accdeps);
MNT_VNODE_FOREACH_ALL(vp, mp, mvp) {
/*
2012-03-28 13:47:07 +00:00
* Depend on the vnode interlock to keep things stable enough
* for a quick test. Since there might be hundreds of
* thousands of vnodes, we cannot afford even a subroutine
* call unless there's a good chance that we have work to do.
*/
if (vp->v_type == VNON) {
VI_UNLOCK(vp);
continue;
}
1994-05-24 10:09:53 +00:00
ip = VTOI(vp);
if ((ip->i_flag &
(IN_ACCESS | IN_CHANGE | IN_MODIFIED | IN_UPDATE)) == 0 &&
vp->v_bufobj.bo_dirty.bv_cnt == 0) {
VI_UNLOCK(vp);
1994-05-24 10:09:53 +00:00
continue;
}
if ((error = vget(vp, lockreq, td)) != 0) {
if (error == ENOENT || error == ENOLCK) {
MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
goto loop;
}
continue;
}
if ((error = ffs_syncvnode(vp, waitfor, 0)) != 0)
allerror = error;
vput(vp);
1994-05-24 10:09:53 +00:00
}
/*
2002-05-16 21:28:32 +00:00
* Force stale filesystem control information to be flushed.
1994-05-24 10:09:53 +00:00
*/
This patch corrects the first round of panics and hangs reported with the new snapshot code. Update addaliasu to correctly implement the semantics of the old checkalias function. When a device vnode first comes into existence, check to see if an anonymous vnode for the same device was created at boot time by bdevvp(). If so, adopt the bdevvp vnode rather than creating a new vnode for the device. This corrects a problem which caused the kernel to panic when taking a snapshot of the root filesystem. Change the calling convention of vn_write_suspend_wait() to be the same as vn_start_write(). Split out softdep_flushworklist() from softdep_flushfiles() so that it can be used to clear the work queue when suspending filesystem operations. Access to buffers becomes recursive so that snapshots can recursively traverse their indirect blocks using ffs_copyonwrite() when checking for the need for copy on write when flushing one of their own indirect blocks. This eliminates a deadlock between the syncer daemon and a process taking a snapshot. Ensure that softdep_process_worklist() can never block because of a snapshot being taken. This eliminates a problem with buffer starvation. Cleanup change in ffs_sync() which did not synchronously wait when MNT_WAIT was specified. The result was an unclean filesystem panic when doing forcible unmount with heavy filesystem I/O in progress. Return a zero'ed block when reading a block that was not in use at the time that a snapshot was taken. Normally, these blocks should never be read. However, the readahead code will occationally read them which can cause unexpected behavior. Clean up the debugging code that ensures that no blocks be written on a filesystem while it is suspended. Snapshots must explicitly label the blocks that they are writing during the suspension so that they do not cause a `write on suspended filesystem' panic. Reorganize ffs_copyonwrite() to eliminate a deadlock and also to prevent a race condition that would permit the same block to be copied twice. This change eliminates an unexpected soft updates inconsistency in fsck caused by the double allocation. Use bqrelse rather than brelse for buffers that will be needed soon again by the snapshot code. This improves snapshot performance.
2000-07-24 05:28:33 +00:00
if (waitfor == MNT_WAIT) {
if ((error = softdep_flushworklist(ump->um_mountp, &count, td)))
This patch corrects the first round of panics and hangs reported with the new snapshot code. Update addaliasu to correctly implement the semantics of the old checkalias function. When a device vnode first comes into existence, check to see if an anonymous vnode for the same device was created at boot time by bdevvp(). If so, adopt the bdevvp vnode rather than creating a new vnode for the device. This corrects a problem which caused the kernel to panic when taking a snapshot of the root filesystem. Change the calling convention of vn_write_suspend_wait() to be the same as vn_start_write(). Split out softdep_flushworklist() from softdep_flushfiles() so that it can be used to clear the work queue when suspending filesystem operations. Access to buffers becomes recursive so that snapshots can recursively traverse their indirect blocks using ffs_copyonwrite() when checking for the need for copy on write when flushing one of their own indirect blocks. This eliminates a deadlock between the syncer daemon and a process taking a snapshot. Ensure that softdep_process_worklist() can never block because of a snapshot being taken. This eliminates a problem with buffer starvation. Cleanup change in ffs_sync() which did not synchronously wait when MNT_WAIT was specified. The result was an unclean filesystem panic when doing forcible unmount with heavy filesystem I/O in progress. Return a zero'ed block when reading a block that was not in use at the time that a snapshot was taken. Normally, these blocks should never be read. However, the readahead code will occationally read them which can cause unexpected behavior. Clean up the debugging code that ensures that no blocks be written on a filesystem while it is suspended. Snapshots must explicitly label the blocks that they are writing during the suspension so that they do not cause a `write on suspended filesystem' panic. Reorganize ffs_copyonwrite() to eliminate a deadlock and also to prevent a race condition that would permit the same block to be copied twice. This change eliminates an unexpected soft updates inconsistency in fsck caused by the double allocation. Use bqrelse rather than brelse for buffers that will be needed soon again by the snapshot code. This improves snapshot performance.
2000-07-24 05:28:33 +00:00
allerror = error;
/* Flushed work items may create new vnodes to clean */
if (allerror == 0 && count)
This patch corrects the first round of panics and hangs reported with the new snapshot code. Update addaliasu to correctly implement the semantics of the old checkalias function. When a device vnode first comes into existence, check to see if an anonymous vnode for the same device was created at boot time by bdevvp(). If so, adopt the bdevvp vnode rather than creating a new vnode for the device. This corrects a problem which caused the kernel to panic when taking a snapshot of the root filesystem. Change the calling convention of vn_write_suspend_wait() to be the same as vn_start_write(). Split out softdep_flushworklist() from softdep_flushfiles() so that it can be used to clear the work queue when suspending filesystem operations. Access to buffers becomes recursive so that snapshots can recursively traverse their indirect blocks using ffs_copyonwrite() when checking for the need for copy on write when flushing one of their own indirect blocks. This eliminates a deadlock between the syncer daemon and a process taking a snapshot. Ensure that softdep_process_worklist() can never block because of a snapshot being taken. This eliminates a problem with buffer starvation. Cleanup change in ffs_sync() which did not synchronously wait when MNT_WAIT was specified. The result was an unclean filesystem panic when doing forcible unmount with heavy filesystem I/O in progress. Return a zero'ed block when reading a block that was not in use at the time that a snapshot was taken. Normally, these blocks should never be read. However, the readahead code will occationally read them which can cause unexpected behavior. Clean up the debugging code that ensures that no blocks be written on a filesystem while it is suspended. Snapshots must explicitly label the blocks that they are writing during the suspension so that they do not cause a `write on suspended filesystem' panic. Reorganize ffs_copyonwrite() to eliminate a deadlock and also to prevent a race condition that would permit the same block to be copied twice. This change eliminates an unexpected soft updates inconsistency in fsck caused by the double allocation. Use bqrelse rather than brelse for buffers that will be needed soon again by the snapshot code. This improves snapshot performance.
2000-07-24 05:28:33 +00:00
goto loop;
}
#ifdef QUOTA
qsync(mp);
#endif
In the original days of BSD, a sync was issued on every filesystem every 30 seconds. This spike in I/O caused the system to pause every 30 seconds which was quite annoying. So, the way that sync worked was changed so that when a vnode was first dirtied, it was put on a 30-second cleaning queue (see the syncer_workitem_pending queues in kern/vfs_subr.c). If the file has not been written or deleted after 30 seconds, the syncer pushes it out. As the syncer runs once per second, dirty files are trickled out slowly over the 30-second period instead of all at once by a call to sync(2). The one drawback to this is that it does not cover the filesystem metadata. To handle the metadata, vfs_allocate_syncvnode() is called to create a "filesystem syncer vnode" at mount time which cycles around the cleaning queue being sync'ed every 30 seconds. In the original design, the only things it would sync for UFS were the filesystem metadata: inode blocks, cylinder group bitmaps, and the superblock (e.g., by VOP_FSYNC'ing devvp, the device vnode from which the filesystem is mounted). Somewhere in its path to integration with FreeBSD the flushing of the filesystem syncer vnode got changed to sync every vnode associated with the filesystem. The result of this change is to return to the old filesystem-wide flush every 30-seconds behavior and makes the whole 30-second delay per vnode useless. This change goes back to the originally intended trickle out sync behavior. Key to ensuring that all the intended semantics are preserved (e.g., that all inode updates get flushed within a bounded period of time) is that all inode modifications get pushed to their corresponding inode blocks so that the metadata flush by the filesystem syncer vnode gets them to the disk in a timely way. Thanks to Konstantin Belousov (kib@) for doing the audit and commit -r231122 which ensures that all of these updates are being made. Reviewed by: kib Tested by: scottl MFC after: 2 weeks
2012-02-07 20:43:28 +00:00
devvp = ump->um_devvp;
bo = &devvp->v_bufobj;
BO_LOCK(bo);
In the original days of BSD, a sync was issued on every filesystem every 30 seconds. This spike in I/O caused the system to pause every 30 seconds which was quite annoying. So, the way that sync worked was changed so that when a vnode was first dirtied, it was put on a 30-second cleaning queue (see the syncer_workitem_pending queues in kern/vfs_subr.c). If the file has not been written or deleted after 30 seconds, the syncer pushes it out. As the syncer runs once per second, dirty files are trickled out slowly over the 30-second period instead of all at once by a call to sync(2). The one drawback to this is that it does not cover the filesystem metadata. To handle the metadata, vfs_allocate_syncvnode() is called to create a "filesystem syncer vnode" at mount time which cycles around the cleaning queue being sync'ed every 30 seconds. In the original design, the only things it would sync for UFS were the filesystem metadata: inode blocks, cylinder group bitmaps, and the superblock (e.g., by VOP_FSYNC'ing devvp, the device vnode from which the filesystem is mounted). Somewhere in its path to integration with FreeBSD the flushing of the filesystem syncer vnode got changed to sync every vnode associated with the filesystem. The result of this change is to return to the old filesystem-wide flush every 30-seconds behavior and makes the whole 30-second delay per vnode useless. This change goes back to the originally intended trickle out sync behavior. Key to ensuring that all the intended semantics are preserved (e.g., that all inode updates get flushed within a bounded period of time) is that all inode modifications get pushed to their corresponding inode blocks so that the metadata flush by the filesystem syncer vnode gets them to the disk in a timely way. Thanks to Konstantin Belousov (kib@) for doing the audit and commit -r231122 which ensures that all of these updates are being made. Reviewed by: kib Tested by: scottl MFC after: 2 weeks
2012-02-07 20:43:28 +00:00
if (bo->bo_numoutput > 0 || bo->bo_dirty.bv_cnt > 0) {
BO_UNLOCK(bo);
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
if ((error = VOP_FSYNC(devvp, waitfor, td)) != 0)
allerror = error;
VOP_UNLOCK(devvp, 0);
if (allerror == 0 && waitfor == MNT_WAIT)
goto loop;
} else if (suspend != 0) {
if (softdep_check_suspend(mp,
devvp,
softdep_deps,
softdep_accdeps,
secondary_writes,
secondary_accwrites) != 0) {
MNT_IUNLOCK(mp);
goto loop; /* More work needed */
}
mtx_assert(MNT_MTX(mp), MA_OWNED);
mp->mnt_kern_flag |= MNTK_SUSPEND2 | MNTK_SUSPENDED;
MNT_IUNLOCK(mp);
suspended = 1;
} else
BO_UNLOCK(bo);
/*
* Write back modified superblock.
*/
if (fs->fs_fmod != 0 &&
(error = ffs_sbupdate(ump, waitfor, suspended)) != 0)
allerror = error;
1994-05-24 10:09:53 +00:00
return (allerror);
}
int
ffs_vget(mp, ino, flags, vpp)
1994-05-24 10:09:53 +00:00
struct mount *mp;
ino_t ino;
int flags;
1994-05-24 10:09:53 +00:00
struct vnode **vpp;
{
return (ffs_vgetf(mp, ino, flags, vpp, 0));
}
int
ffs_vgetf(mp, ino, flags, vpp, ffs_flags)
struct mount *mp;
ino_t ino;
int flags;
struct vnode **vpp;
int ffs_flags;
1994-05-24 10:09:53 +00:00
{
struct fs *fs;
struct inode *ip;
1994-05-24 10:09:53 +00:00
struct ufsmount *ump;
struct buf *bp;
struct vnode *vp;
struct cdev *dev;
int error;
1994-05-24 10:09:53 +00:00
error = vfs_hash_get(mp, ino, flags, curthread, vpp, NULL, NULL);
if (error || *vpp != NULL)
2005-03-14 10:21:16 +00:00
return (error);
/*
* We must promote to an exclusive lock for vnode creation. This
* can happen if lookup is passed LOCKSHARED.
*/
if ((flags & LK_TYPE_MASK) == LK_SHARED) {
flags &= ~LK_TYPE_MASK;
flags |= LK_EXCLUSIVE;
}
/*
* We do not lock vnode creation as it is believed to be too
* expensive for such rare case as simultaneous creation of vnode
* for same ino by different processes. We just allow them to race
* and check later to decide who wins. Let the race begin!
*/
2005-03-14 10:21:16 +00:00
ump = VFSTOUFS(mp);
dev = ump->um_dev;
fs = ump->um_fs;
1994-05-24 10:09:53 +00:00
/*
* If this malloc() is performed after the getnewvnode()
* it might block, leaving a vnode with a NULL v_data to be
* found by ffs_sync() if a sync happens to fire right then,
* which will cause a panic because ffs_sync() blindly
* dereferences vp->v_data (as well it should).
*/
2005-03-14 10:21:16 +00:00
ip = uma_zalloc(uma_inode, M_WAITOK | M_ZERO);
1994-05-24 10:09:53 +00:00
/* Allocate a new vnode/inode. */
if (fs->fs_magic == FS_UFS1_MAGIC)
error = getnewvnode("ufs", mp, &ffs_vnodeops1, &vp);
else
error = getnewvnode("ufs", mp, &ffs_vnodeops2, &vp);
if (error) {
1994-05-24 10:09:53 +00:00
*vpp = NULL;
uma_zfree(uma_inode, ip);
1994-05-24 10:09:53 +00:00
return (error);
}
/*
* FFS supports recursive locking.
*/
lockmgr(vp->v_vnlock, LK_EXCLUSIVE, NULL);
VN_LOCK_AREC(vp);
1994-05-24 10:09:53 +00:00
vp->v_data = ip;
vp->v_bufobj.bo_bsize = fs->fs_bsize;
1994-05-24 10:09:53 +00:00
ip->i_vnode = vp;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
ip->i_ump = ump;
2004-07-07 20:04:06 +00:00
ip->i_fs = fs;
1994-05-24 10:09:53 +00:00
ip->i_dev = dev;
ip->i_number = ino;
ip->i_ea_refs = 0;
1994-05-24 10:09:53 +00:00
#ifdef QUOTA
{
int i;
for (i = 0; i < MAXQUOTAS; i++)
ip->i_dquot[i] = NODQUOT;
}
1994-05-24 10:09:53 +00:00
#endif
if (ffs_flags & FFSV_FORCEINSMQ)
vp->v_vflag |= VV_FORCEINSMQ;
error = insmntque(vp, mp);
if (error != 0) {
uma_zfree(uma_inode, ip);
*vpp = NULL;
return (error);
}
vp->v_vflag &= ~VV_FORCEINSMQ;
error = vfs_hash_insert(vp, ino, flags, curthread, vpp, NULL, NULL);
if (error || *vpp != NULL)
return (error);
1994-05-24 10:09:53 +00:00
/* Read in the disk contents for the inode, copy into the inode. */
error = bread(ump->um_devvp, fsbtodb(fs, ino_to_fsba(fs, ino)),
(int)fs->fs_bsize, NOCRED, &bp);
if (error) {
1994-05-24 10:09:53 +00:00
/*
* The inode does not contain anything useful, so it would
* be misleading to leave it on its hash chain. With mode
* still zero, it will be unlinked and returned to the free
* list by vput().
*/
brelse(bp);
vput(vp);
1994-05-24 10:09:53 +00:00
*vpp = NULL;
return (error);
}
if (ip->i_ump->um_fstype == UFS1)
ip->i_din1 = uma_zalloc(uma_ufs1, M_WAITOK);
else
ip->i_din2 = uma_zalloc(uma_ufs2, M_WAITOK);
ffs_load_inode(bp, ip, fs, ino);
if (DOINGSOFTDEP(vp))
softdep_load_inodeblock(ip);
else
ip->i_effnlink = ip->i_nlink;
bqrelse(bp);
1994-05-24 10:09:53 +00:00
/*
* Initialize the vnode from the inode, check for aliases.
* Note that the underlying vnode may have changed.
*/
if (ip->i_ump->um_fstype == UFS1)
error = ufs_vinit(mp, &ffs_fifoops1, &vp);
else
error = ufs_vinit(mp, &ffs_fifoops2, &vp);
if (error) {
1994-05-24 10:09:53 +00:00
vput(vp);
*vpp = NULL;
return (error);
}
1994-05-24 10:09:53 +00:00
/*
* Finish inode initialization.
1994-05-24 10:09:53 +00:00
*/
if (vp->v_type != VFIFO) {
/* FFS supports shared locking for all files except fifos. */
VN_LOCK_ASHARE(vp);
}
1994-05-24 10:09:53 +00:00
/*
* Set up a generation number for this inode if it does not
* already have one. This should only happen on old filesystems.
*/
if (ip->i_gen == 0) {
ip->i_gen = arc4random() / 2 + 1;
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
if ((vp->v_mount->mnt_flag & MNT_RDONLY) == 0) {
1994-05-24 10:09:53 +00:00
ip->i_flag |= IN_MODIFIED;
DIP_SET(ip, i_gen, ip->i_gen);
This commit adds basic support for the UFS2 filesystem. The UFS2 filesystem expands the inode to 256 bytes to make space for 64-bit block pointers. It also adds a file-creation time field, an ability to use jumbo blocks per inode to allow extent like pointer density, and space for extended attributes (up to twice the filesystem block size worth of attributes, e.g., on a 16K filesystem, there is space for 32K of attributes). UFS2 fully supports and runs existing UFS1 filesystems. New filesystems built using newfs can be built in either UFS1 or UFS2 format using the -O option. In this commit UFS1 is the default format, so if you want to build UFS2 format filesystems, you must specify -O 2. This default will be changed to UFS2 when UFS2 proves itself to be stable. In this commit the boot code for reading UFS2 filesystems is not compiled (see /sys/boot/common/ufsread.c) as there is insufficient space in the boot block. Once the size of the boot block is increased, this code can be defined. Things to note: the definition of SBSIZE has changed to SBLOCKSIZE. The header file <ufs/ufs/dinode.h> must be included before <ufs/ffs/fs.h> so as to get the definitions of ufs2_daddr_t and ufs_lbn_t. Still TODO: Verify that the first level bootstraps work for all the architectures. Convert the utility ffsinfo to understand UFS2 and test growfs. Add support for the extended attribute storage. Update soft updates to ensure integrity of extended attribute storage. Switch the current extended attribute interfaces to use the extended attribute storage. Add the extent like functionality (framework is there, but is currently never used). Sponsored by: DARPA & NAI Labs. Reviewed by: Poul-Henning Kamp <phk@freebsd.org>
2002-06-21 06:18:05 +00:00
}
1994-05-24 10:09:53 +00:00
}
#ifdef MAC
if ((mp->mnt_flag & MNT_MULTILABEL) && ip->i_mode) {
/*
* If this vnode is already allocated, and we're running
* multi-label, attempt to perform a label association
* from the extended attributes on the inode.
*/
error = mac_vnode_associate_extattr(mp, vp);
if (error) {
/* ufs_inactive will release ip->i_devvp ref. */
vput(vp);
*vpp = NULL;
return (error);
}
}
#endif
1994-05-24 10:09:53 +00:00
*vpp = vp;
return (0);
}
/*
* File handle to vnode
*
* Have to be really careful about stale file handles:
* - check that the inode number is valid
* - call ffs_vget() to get the locked inode
* - check for an unallocated inode (i_mode == 0)
* - check that the given client host has export rights and return
* those rights via. exflagsp and credanonp
*/
static int
ffs_fhtovp(mp, fhp, flags, vpp)
struct mount *mp;
1994-05-24 10:09:53 +00:00
struct fid *fhp;
int flags;
1994-05-24 10:09:53 +00:00
struct vnode **vpp;
{
struct ufid *ufhp;
1994-05-24 10:09:53 +00:00
struct fs *fs;
ufhp = (struct ufid *)fhp;
fs = VFSTOUFS(mp)->um_fs;
if (ufhp->ufid_ino < ROOTINO ||
ufhp->ufid_ino >= fs->fs_ncg * fs->fs_ipg)
return (ESTALE);
return (ufs_fhtovp(mp, ufhp, flags, vpp));
1994-05-24 10:09:53 +00:00
}
/*
* Initialize the filesystem.
*/
static int
ffs_init(vfsp)
struct vfsconf *vfsp;
{
softdep_initialize();
return (ufs_init(vfsp));
}
/*
* Undo the work of ffs_init().
*/
static int
ffs_uninit(vfsp)
struct vfsconf *vfsp;
{
int ret;
ret = ufs_uninit(vfsp);
softdep_uninitialize();
return (ret);
}
1994-05-24 10:09:53 +00:00
/*
* Write a superblock and associated information back to disk.
*/
int
ffs_sbupdate(ump, waitfor, suspended)
struct ufsmount *ump;
1994-05-24 10:09:53 +00:00
int waitfor;
int suspended;
1994-05-24 10:09:53 +00:00
{
struct fs *fs = ump->um_fs;
struct buf *sbbp;
struct buf *bp;
1994-05-24 10:09:53 +00:00
int blks;
void *space;
int i, size, error, allerror = 0;
1994-05-24 10:09:53 +00:00
if (fs->fs_ronly == 1 &&
(ump->um_mountp->mnt_flag & (MNT_RDONLY | MNT_UPDATE)) !=
(MNT_RDONLY | MNT_UPDATE) && ump->um_fsckpid == 0)
panic("ffs_sbupdate: write read-only filesystem");
/*
* We use the superblock's buf to serialize calls to ffs_sbupdate().
*/
sbbp = getblk(ump->um_devvp, btodb(fs->fs_sblockloc),
(int)fs->fs_sbsize, 0, 0, 0);
/*
* First write back the summary information.
*/
1994-05-24 10:09:53 +00:00
blks = howmany(fs->fs_cssize, fs->fs_fsize);
space = fs->fs_csp;
1994-05-24 10:09:53 +00:00
for (i = 0; i < blks; i += fs->fs_frag) {
size = fs->fs_bsize;
if (i + fs->fs_frag > blks)
size = (blks - i) * fs->fs_fsize;
bp = getblk(ump->um_devvp, fsbtodb(fs, fs->fs_csaddr + i),
size, 0, 0, 0);
1994-05-24 10:09:53 +00:00
bcopy(space, bp->b_data, (u_int)size);
space = (char *)space + size;
if (suspended)
bp->b_flags |= B_VALIDSUSPWRT;
if (waitfor != MNT_WAIT)
1994-05-24 10:09:53 +00:00
bawrite(bp);
else if ((error = bwrite(bp)) != 0)
allerror = error;
1994-05-24 10:09:53 +00:00
}
/*
* Now write back the superblock itself. If any errors occurred
* up to this point, then fail so that the superblock avoids
* being written out as clean.
*/
if (allerror) {
brelse(sbbp);
return (allerror);
}
bp = sbbp;
if (fs->fs_magic == FS_UFS1_MAGIC && fs->fs_sblockloc != SBLOCK_UFS1 &&
(fs->fs_flags & FS_FLAGS_UPDATED) == 0) {
printf("WARNING: %s: correcting fs_sblockloc from %jd to %d\n",
fs->fs_fsmnt, fs->fs_sblockloc, SBLOCK_UFS1);
fs->fs_sblockloc = SBLOCK_UFS1;
}
if (fs->fs_magic == FS_UFS2_MAGIC && fs->fs_sblockloc != SBLOCK_UFS2 &&
(fs->fs_flags & FS_FLAGS_UPDATED) == 0) {
printf("WARNING: %s: correcting fs_sblockloc from %jd to %d\n",
fs->fs_fsmnt, fs->fs_sblockloc, SBLOCK_UFS2);
fs->fs_sblockloc = SBLOCK_UFS2;
}
fs->fs_fmod = 0;
fs->fs_time = time_second;
if (fs->fs_flags & FS_DOSOFTDEP)
softdep_setup_sbupdate(ump, (struct fs *)bp->b_data, bp);
bcopy((caddr_t)fs, bp->b_data, (u_int)fs->fs_sbsize);
ffs_oldfscompat_write((struct fs *)bp->b_data, ump);
if (suspended)
bp->b_flags |= B_VALIDSUSPWRT;
if (waitfor != MNT_WAIT)
bawrite(bp);
else if ((error = bwrite(bp)) != 0)
allerror = error;
return (allerror);
1994-05-24 10:09:53 +00:00
}
static int
ffs_extattrctl(struct mount *mp, int cmd, struct vnode *filename_vp,
int attrnamespace, const char *attrname)
{
#ifdef UFS_EXTATTR
return (ufs_extattrctl(mp, cmd, filename_vp, attrnamespace,
attrname));
#else
return (vfs_stdextattrctl(mp, cmd, filename_vp, attrnamespace,
attrname));
#endif
}
static void
ffs_ifree(struct ufsmount *ump, struct inode *ip)
{
if (ump->um_fstype == UFS1 && ip->i_din1 != NULL)
uma_zfree(uma_ufs1, ip->i_din1);
else if (ip->i_din2 != NULL)
uma_zfree(uma_ufs2, ip->i_din2);
uma_zfree(uma_inode, ip);
}
static int dobkgrdwrite = 1;
SYSCTL_INT(_debug, OID_AUTO, dobkgrdwrite, CTLFLAG_RW, &dobkgrdwrite, 0,
"Do background writes (honoring the BV_BKGRDWRITE flag)?");
/*
* Complete a background write started from bwrite.
*/
static void
ffs_backgroundwritedone(struct buf *bp)
{
struct bufobj *bufobj;
struct buf *origbp;
/*
* Find the original buffer that we are writing.
*/
bufobj = bp->b_bufobj;
BO_LOCK(bufobj);
if ((origbp = gbincore(bp->b_bufobj, bp->b_lblkno)) == NULL)
panic("backgroundwritedone: lost buffer");
/* Grab an extra reference to be dropped by the bufdone() below. */
bufobj_wrefl(bufobj);
BO_UNLOCK(bufobj);
/*
* Process dependencies then return any unfinished ones.
*/
2007-04-04 07:29:53 +00:00
if (!LIST_EMPTY(&bp->b_dep))
buf_complete(bp);
#ifdef SOFTUPDATES
2007-04-04 07:29:53 +00:00
if (!LIST_EMPTY(&bp->b_dep))
softdep_move_dependencies(bp, origbp);
#endif
/*
* This buffer is marked B_NOCACHE so when it is released
* by biodone it will be tossed.
*/
bp->b_flags |= B_NOCACHE;
bp->b_flags &= ~B_CACHE;
bufdone(bp);
BO_LOCK(bufobj);
/*
* Clear the BV_BKGRDINPROG flag in the original buffer
* and awaken it if it is waiting for the write to complete.
* If BV_BKGRDINPROG is not set in the original buffer it must
* have been released and re-instantiated - which is not legal.
*/
KASSERT((origbp->b_vflags & BV_BKGRDINPROG),
("backgroundwritedone: lost buffer2"));
origbp->b_vflags &= ~BV_BKGRDINPROG;
if (origbp->b_vflags & BV_BKGRDWAIT) {
origbp->b_vflags &= ~BV_BKGRDWAIT;
wakeup(&origbp->b_xflags);
}
BO_UNLOCK(bufobj);
}
/*
* Write, release buffer on completion. (Done by iodone
* if async). Do not bother writing anything if the buffer
* is invalid.
*
* Note that we set B_CACHE here, indicating that buffer is
* fully valid and thus cacheable. This is true even of NFS
2010-09-17 09:14:40 +00:00
* now so we set it generally. This could be set either here
* or in biodone() since the I/O is synchronous. We put it
* here.
*/
static int
ffs_bufwrite(struct buf *bp)
{
int oldflags, s;
struct buf *newbp;
CTR3(KTR_BUF, "bufwrite(%p) vp %p flags %X", bp, bp->b_vp, bp->b_flags);
if (bp->b_flags & B_INVAL) {
brelse(bp);
return (0);
}
oldflags = bp->b_flags;
if (!BUF_ISLOCKED(bp))
panic("bufwrite: buffer is not busy???");
s = splbio();
/*
* If a background write is already in progress, delay
* writing this block if it is asynchronous. Otherwise
* wait for the background write to complete.
*/
BO_LOCK(bp->b_bufobj);
if (bp->b_vflags & BV_BKGRDINPROG) {
if (bp->b_flags & B_ASYNC) {
BO_UNLOCK(bp->b_bufobj);
splx(s);
bdwrite(bp);
return (0);
}
bp->b_vflags |= BV_BKGRDWAIT;
msleep(&bp->b_xflags, BO_MTX(bp->b_bufobj), PRIBIO, "bwrbg", 0);
if (bp->b_vflags & BV_BKGRDINPROG)
panic("bufwrite: still writing");
}
BO_UNLOCK(bp->b_bufobj);
/*
* If this buffer is marked for background writing and we
* do not have to wait for it, make a copy and write the
* copy so as to leave this buffer ready for further use.
*
* This optimization eats a lot of memory. If we have a page
* or buffer shortfall we can't do it.
*/
2010-09-17 09:14:40 +00:00
if (dobkgrdwrite && (bp->b_xflags & BX_BKGRDWRITE) &&
(bp->b_flags & B_ASYNC) &&
!vm_page_count_severe() &&
!buf_dirty_count_severe()) {
KASSERT(bp->b_iodone == NULL,
("bufwrite: needs chained iodone (%p)", bp->b_iodone));
/* get a new block */
Fix two issues with bufdaemon, often causing the processes to hang in the "nbufkv" sleep. First, ffs background cg group block write requests a new buffer for the shadow copy. When ffs_bufwrite() is called from the bufdaemon due to buffers shortage, requesting the buffer deadlock bufdaemon. Introduce a new flag for getnewbuf(), GB_NOWAIT_BD, to request getblk to not block while allocating the buffer, and return failure instead. Add a flag argument to the geteblk to allow to pass the flags to getblk(). Do not repeat the getnewbuf() call from geteblk if buffer allocation failed and either GB_NOWAIT_BD is specified, or geteblk() is called from bufdaemon (or its helper, see below). In ffs_bufwrite(), fall back to synchronous cg block write if shadow block allocation failed. Since r107847, buffer write assumes that vnode owning the buffer is locked. The second problem is that buffer cache may accumulate many buffers belonging to limited number of vnodes. With such workload, quite often threads that own the mentioned vnodes locks are trying to read another block from the vnodes, and, due to buffer cache exhaustion, are asking bufdaemon for help. Bufdaemon is unable to make any substantial progress because the vnodes are locked. Allow the threads owning vnode locks to help the bufdaemon by doing the flush pass over the buffer cache before getnewbuf() is going to uninterruptible sleep. Move the flushing code from buf_daemon() to new helper function buf_do_flush(), that is called from getnewbuf(). The number of buffers flushed by single call to buf_do_flush() from getnewbuf() is limited by new sysctl vfs.flushbufqtarget. Prevent recursive calls to buf_do_flush() by marking the bufdaemon and threads that temporarily help bufdaemon by TDP_BUFNEED flag. In collaboration with: pho Reviewed by: tegge (previous version) Tested by: glebius, yandex ... MFC after: 3 weeks
2009-03-16 15:39:46 +00:00
newbp = geteblk(bp->b_bufsize, GB_NOWAIT_BD);
if (newbp == NULL)
goto normal_write;
/*
* set it to be identical to the old block. We have to
* set b_lblkno and BKGRDMARKER before calling bgetvp()
* to avoid confusing the splay tree and gbincore().
*/
memcpy(newbp->b_data, bp->b_data, bp->b_bufsize);
newbp->b_lblkno = bp->b_lblkno;
newbp->b_xflags |= BX_BKGRDMARKER;
BO_LOCK(bp->b_bufobj);
bp->b_vflags |= BV_BKGRDINPROG;
bgetvp(bp->b_vp, newbp);
BO_UNLOCK(bp->b_bufobj);
newbp->b_bufobj = &bp->b_vp->v_bufobj;
newbp->b_blkno = bp->b_blkno;
newbp->b_offset = bp->b_offset;
newbp->b_iodone = ffs_backgroundwritedone;
newbp->b_flags |= B_ASYNC;
newbp->b_flags &= ~B_INVAL;
#ifdef SOFTUPDATES
/*
* Move over the dependencies. If there are rollbacks,
* leave the parent buffer dirtied as it will need to
* be written again.
*/
if (LIST_EMPTY(&bp->b_dep) ||
softdep_move_dependencies(bp, newbp) == 0)
bundirty(bp);
#else
bundirty(bp);
2010-09-17 09:14:40 +00:00
#endif
/*
* Initiate write on the copy, release the original to
* the B_LOCKED queue so that it cannot go away until
* the background write completes. If not locked it could go
* away and then be reconstituted while it was being written.
* If the reconstituted buffer were written, we could end up
* with two background copies being written at the same time.
*/
bqrelse(bp);
bp = newbp;
} else
/* Mark the buffer clean */
bundirty(bp);
/* Let the normal bufwrite do the rest for us */
Fix two issues with bufdaemon, often causing the processes to hang in the "nbufkv" sleep. First, ffs background cg group block write requests a new buffer for the shadow copy. When ffs_bufwrite() is called from the bufdaemon due to buffers shortage, requesting the buffer deadlock bufdaemon. Introduce a new flag for getnewbuf(), GB_NOWAIT_BD, to request getblk to not block while allocating the buffer, and return failure instead. Add a flag argument to the geteblk to allow to pass the flags to getblk(). Do not repeat the getnewbuf() call from geteblk if buffer allocation failed and either GB_NOWAIT_BD is specified, or geteblk() is called from bufdaemon (or its helper, see below). In ffs_bufwrite(), fall back to synchronous cg block write if shadow block allocation failed. Since r107847, buffer write assumes that vnode owning the buffer is locked. The second problem is that buffer cache may accumulate many buffers belonging to limited number of vnodes. With such workload, quite often threads that own the mentioned vnodes locks are trying to read another block from the vnodes, and, due to buffer cache exhaustion, are asking bufdaemon for help. Bufdaemon is unable to make any substantial progress because the vnodes are locked. Allow the threads owning vnode locks to help the bufdaemon by doing the flush pass over the buffer cache before getnewbuf() is going to uninterruptible sleep. Move the flushing code from buf_daemon() to new helper function buf_do_flush(), that is called from getnewbuf(). The number of buffers flushed by single call to buf_do_flush() from getnewbuf() is limited by new sysctl vfs.flushbufqtarget. Prevent recursive calls to buf_do_flush() by marking the bufdaemon and threads that temporarily help bufdaemon by TDP_BUFNEED flag. In collaboration with: pho Reviewed by: tegge (previous version) Tested by: glebius, yandex ... MFC after: 3 weeks
2009-03-16 15:39:46 +00:00
normal_write:
return (bufwrite(bp));
}
static void
ffs_geom_strategy(struct bufobj *bo, struct buf *bp)
{
struct vnode *vp;
int error;
struct buf *tbp;
int nocopy;
vp = bo->__bo_vnode;
if (bp->b_iocmd == BIO_WRITE) {
if ((bp->b_flags & B_VALIDSUSPWRT) == 0 &&
bp->b_vp != NULL && bp->b_vp->v_mount != NULL &&
(bp->b_vp->v_mount->mnt_kern_flag & MNTK_SUSPENDED) != 0)
panic("ffs_geom_strategy: bad I/O");
nocopy = bp->b_flags & B_NOCOPY;
bp->b_flags &= ~(B_VALIDSUSPWRT | B_NOCOPY);
if ((vp->v_vflag & VV_COPYONWRITE) && nocopy == 0 &&
vp->v_rdev->si_snapdata != NULL) {
if ((bp->b_flags & B_CLUSTER) != 0) {
runningbufwakeup(bp);
TAILQ_FOREACH(tbp, &bp->b_cluster.cluster_head,
b_cluster.cluster_entry) {
error = ffs_copyonwrite(vp, tbp);
if (error != 0 &&
error != EOPNOTSUPP) {
bp->b_error = error;
bp->b_ioflags |= BIO_ERROR;
bufdone(bp);
return;
}
}
bp->b_runningbufspace = bp->b_bufsize;
atomic_add_long(&runningbufspace,
bp->b_runningbufspace);
} else {
error = ffs_copyonwrite(vp, bp);
if (error != 0 && error != EOPNOTSUPP) {
bp->b_error = error;
bp->b_ioflags |= BIO_ERROR;
bufdone(bp);
return;
}
}
}
#ifdef SOFTUPDATES
if ((bp->b_flags & B_CLUSTER) != 0) {
TAILQ_FOREACH(tbp, &bp->b_cluster.cluster_head,
b_cluster.cluster_entry) {
2007-04-04 07:29:53 +00:00
if (!LIST_EMPTY(&tbp->b_dep))
buf_start(tbp);
}
} else {
2007-04-04 07:29:53 +00:00
if (!LIST_EMPTY(&bp->b_dep))
buf_start(bp);
}
#endif
}
Move UFS from DEVFS backing to GEOM backing. This eliminates a bunch of vnode overhead (approx 1-2 % speed improvement) and gives us more control over the access to the storage device. Access counts on the underlying device are not correctly tracked and therefore it is possible to read-only mount the same disk device multiple times: syv# mount -p /dev/md0 /var ufs rw 2 2 /dev/ad0 /mnt ufs ro 1 1 /dev/ad0 /mnt2 ufs ro 1 1 /dev/ad0 /mnt3 ufs ro 1 1 Since UFS/FFS is not a synchrousely consistent filesystem (ie: it caches things in RAM) this is not possible with read-write mounts, and the system will correctly reject this. Details: Add a geom consumer and a bufobj pointer to ufsmount. Eliminate the vnode argument from softdep_disk_prewrite(). Pick the vnode out of bp->b_vp for now. Eventually we should find it through bp->b_bufobj->b_private. In the mountcode, use g_vfs_open() once we have used VOP_ACCESS() to check permissions. When upgrading and downgrading between r/o and r/w do the right thing with GEOM access counts. Remove all the workarounds for not being able to do this with VOP_OPEN(). If we are the root mount, drop the exclusive access count until we upgrade to r/w. This allows fsck of the root filesystem and the MNT_RELOAD to work correctly. Set bo_private to the GEOM consumer on the device bufobj. Change the ffs_ops->strategy function to call g_vfs_strategy() In ufs_strategy() directly call the strategy on the disk bufobj. Same in rawread. In ffs_fsync() we will no longer see VCHR device nodes, so remove code which synced the filesystem mounted on it, in case we came there. I'm not sure this code made sense in the first place since we would have taken the specfs route on such a vnode. Redo the highly bogus readblock() function in the snapshot code to something slightly less bogus: Constructing an uio and using physio was really quite a detour. Instead just fill in a bio and ship it down.
2004-10-29 10:15:56 +00:00
g_vfs_strategy(bo, bp);
}
#ifdef DDB
static void
db_print_ffs(struct ufsmount *ump)
{
Implement fully asynchronous partial truncation with softupdates journaling to resolve errors which can cause corruption on recovery with the old synchronous mechanism. - Append partial truncation freework structures to indirdeps while truncation is proceeding. These prevent new block pointers from becoming valid until truncation completes and serialize truncations. - On completion of a partial truncate journal work waits for zeroed pointers to hit indirects. - softdep_journal_freeblocks() handles last frag allocation and last block zeroing. - vtruncbuf/ffs_page_remove moved into softdep_*_freeblocks() so it is only implemented in one place. - Block allocation failure handling moved up one level so it does not proceed with buf locks held. This permits us to do more extensive reclaims when filesystem space is exhausted. - softdep_sync_metadata() is broken into two parts, the first executes once at the start of ffs_syncvnode() and flushes truncations and inode dependencies. The second is called on each locked buf. This eliminates excessive looping and rollbacks. - Improve the mechanism in process_worklist_item() that handles acquiring vnode locks for handle_workitem_remove() so that it works more generally and does not loop excessively over the same worklist items on each call. - Don't corrupt directories by zeroing the tail in fsck. This is only done for regular files. - Push a fsync complete record for files that need it so the checker knows a truncation in the journal is no longer valid. Discussed with: mckusick, kib (ffs_pages_remove and ffs_truncate parts) Tested by: pho
2011-06-10 22:48:35 +00:00
db_printf("mp %p %s devvp %p fs %p su_wl %d su_deps %d su_req %d\n",
ump->um_mountp, ump->um_mountp->mnt_stat.f_mntonname,
ump->um_devvp, ump->um_fs, ump->softdep_on_worklist,
Implement fully asynchronous partial truncation with softupdates journaling to resolve errors which can cause corruption on recovery with the old synchronous mechanism. - Append partial truncation freework structures to indirdeps while truncation is proceeding. These prevent new block pointers from becoming valid until truncation completes and serialize truncations. - On completion of a partial truncate journal work waits for zeroed pointers to hit indirects. - softdep_journal_freeblocks() handles last frag allocation and last block zeroing. - vtruncbuf/ffs_page_remove moved into softdep_*_freeblocks() so it is only implemented in one place. - Block allocation failure handling moved up one level so it does not proceed with buf locks held. This permits us to do more extensive reclaims when filesystem space is exhausted. - softdep_sync_metadata() is broken into two parts, the first executes once at the start of ffs_syncvnode() and flushes truncations and inode dependencies. The second is called on each locked buf. This eliminates excessive looping and rollbacks. - Improve the mechanism in process_worklist_item() that handles acquiring vnode locks for handle_workitem_remove() so that it works more generally and does not loop excessively over the same worklist items on each call. - Don't corrupt directories by zeroing the tail in fsck. This is only done for regular files. - Push a fsync complete record for files that need it so the checker knows a truncation in the journal is no longer valid. Discussed with: mckusick, kib (ffs_pages_remove and ffs_truncate parts) Tested by: pho
2011-06-10 22:48:35 +00:00
ump->softdep_deps, ump->softdep_req);
}
DB_SHOW_COMMAND(ffs, db_show_ffs)
{
struct mount *mp;
struct ufsmount *ump;
if (have_addr) {
ump = VFSTOUFS((struct mount *)addr);
db_print_ffs(ump);
return;
}
TAILQ_FOREACH(mp, &mountlist, mnt_list) {
if (!strcmp(mp->mnt_stat.f_fstypename, ufs_vfsconf.vfc_name))
db_print_ffs(VFSTOUFS(mp));
}
}
#endif /* DDB */