NetBSD/sys/msdosfs/msdosfs_vfsops.c

975 lines
26 KiB
C
Raw Normal View History

1998-03-01 05:20:01 +03:00
/* $NetBSD: msdosfs_vfsops.c,v 1.53 1998/03/01 02:25:11 fvdl Exp $ */
/*-
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``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 TOOLS GMBH 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.
*/
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
* This software is provided "as is".
*
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
* October 1992
*/
1993-12-18 03:50:51 +03:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/namei.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/vnode.h>
#include <miscfs/specfs/specdev.h> /* XXX */ /* defines v_rdev */
#include <sys/mount.h>
#include <sys/buf.h>
#include <sys/file.h>
#include <sys/device.h>
#include <sys/disklabel.h>
#include <sys/ioctl.h>
1993-12-18 03:50:51 +03:00
#include <sys/malloc.h>
1995-10-15 18:34:19 +03:00
#include <sys/dirent.h>
1997-05-08 20:53:06 +04:00
#include <sys/stat.h>
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
1993-12-18 03:50:51 +03:00
#include <msdosfs/bpb.h>
#include <msdosfs/bootsect.h>
#include <msdosfs/direntry.h>
#include <msdosfs/denode.h>
#include <msdosfs/msdosfsmount.h>
#include <msdosfs/fat.h>
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
int msdosfs_mountroot __P((void));
int msdosfs_mount __P((struct mount *, const char *, void *,
struct nameidata *, struct proc *));
1996-02-09 22:13:39 +03:00
int msdosfs_start __P((struct mount *, int, struct proc *));
int msdosfs_unmount __P((struct mount *, int, struct proc *));
int msdosfs_root __P((struct mount *, struct vnode **));
int msdosfs_quotactl __P((struct mount *, int, uid_t, caddr_t, struct proc *));
int msdosfs_statfs __P((struct mount *, struct statfs *, struct proc *));
int msdosfs_sync __P((struct mount *, int, struct ucred *, struct proc *));
int msdosfs_vget __P((struct mount *, ino_t, struct vnode **));
int msdosfs_fhtovp __P((struct mount *, struct fid *, struct mbuf *,
struct vnode **, int *, struct ucred **));
1996-02-09 22:13:39 +03:00
int msdosfs_vptofh __P((struct vnode *, struct fid *));
1998-03-01 05:20:01 +03:00
int msdosfs_sysctl __P((int *, u_int, void *, size_t *, void *, size_t,
struct proc *));
1996-02-09 22:13:39 +03:00
int msdosfs_mountfs __P((struct vnode *, struct mount *, struct proc *,
struct msdosfs_args *));
static int update_mp __P((struct mount *, struct msdosfs_args *));
#define ROOTNAME "root_device"
extern struct vnodeopv_desc msdosfs_vnodeop_opv_desc;
struct vnodeopv_desc *msdosfs_vnodeopv_descs[] = {
&msdosfs_vnodeop_opv_desc,
NULL,
};
struct vfsops msdosfs_vfsops = {
MOUNT_MSDOS,
msdosfs_mount,
msdosfs_start,
msdosfs_unmount,
msdosfs_root,
msdosfs_quotactl,
msdosfs_statfs,
msdosfs_sync,
msdosfs_vget,
msdosfs_fhtovp,
msdosfs_vptofh,
msdosfs_init,
1998-03-01 05:20:01 +03:00
msdosfs_sysctl,
msdosfs_mountroot,
msdosfs_vnodeopv_descs,
};
static int
update_mp(mp, argp)
struct mount *mp;
struct msdosfs_args *argp;
{
struct msdosfsmount *pmp = VFSTOMSDOSFS(mp);
int error;
pmp->pm_gid = argp->gid;
pmp->pm_uid = argp->uid;
pmp->pm_mask = argp->mask & ALLPERMS;
pmp->pm_flags |= argp->flags & MSDOSFSMNT_MNTOPT;
/*
* GEMDOS knows nothing (yet) about win95
*/
if (pmp->pm_flags & MSDOSFSMNT_GEMDOSFS)
pmp->pm_flags |= MSDOSFSMNT_NOWIN95;
if (pmp->pm_flags & MSDOSFSMNT_NOWIN95)
pmp->pm_flags |= MSDOSFSMNT_SHORTNAME;
else if (!(pmp->pm_flags &
(MSDOSFSMNT_SHORTNAME | MSDOSFSMNT_LONGNAME))) {
struct vnode *rootvp;
/*
* Try to divine whether to support Win'95 long filenames
*/
if (FAT32(pmp))
pmp->pm_flags |= MSDOSFSMNT_LONGNAME;
else {
if ((error = msdosfs_root(mp, &rootvp)) != 0)
return error;
pmp->pm_flags |= findwin95(VTODE(rootvp))
? MSDOSFSMNT_LONGNAME
: MSDOSFSMNT_SHORTNAME;
vput(rootvp);
}
}
return 0;
}
int
msdosfs_mountroot()
{
1998-03-01 05:20:01 +03:00
struct mount *mp;
extern struct vnode *rootvp;
struct proc *p = curproc; /* XXX */
int error;
struct msdosfs_args args;
if (root_device->dv_class != DV_DISK)
return (ENODEV);
/*
* Get vnodes for swapdev and rootdev.
*/
if (bdevvp(rootdev, &rootvp))
panic("msdosfs_mountroot: can't setup rootvp");
1998-03-01 05:20:01 +03:00
if ((error = vfs_rootmountalloc(MOUNT_MSDOS, "root_device", &mp)))
return (error);
args.flags = 0;
args.uid = 0;
args.gid = 0;
args.mask = 0777;
if ((error = msdosfs_mountfs(rootvp, mp, p, &args)) != 0) {
1998-03-01 05:20:01 +03:00
mp->mnt_op->vfs_refcount--;
vfs_unbusy(mp);
free(mp, M_MOUNT);
return (error);
}
if ((error = update_mp(mp, &args)) != 0) {
(void)msdosfs_unmount(mp, 0, p);
1998-03-01 05:20:01 +03:00
vfs_unbusy(mp);
free(mp, M_MOUNT);
return (error);
}
1998-03-01 05:20:01 +03:00
simple_lock(&mountlist_slock);
CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list);
1998-03-01 05:20:01 +03:00
simple_unlock(&mountlist_slock);
(void)msdosfs_statfs(mp, &mp->mnt_stat, p);
1998-03-01 05:20:01 +03:00
vfs_unbusy(mp);
return (0);
}
1996-02-09 22:13:39 +03:00
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* mp - path - addr in user space of mount point (ie /usr or whatever)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
* data - addr in user space of mount params including the name of the block
* special file to treat as a filesystem.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
int
msdosfs_mount(mp, path, data, ndp, p)
struct mount *mp;
const char *path;
void *data;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
struct nameidata *ndp;
struct proc *p;
{
struct vnode *devvp; /* vnode for blk device to mount */
struct msdosfs_args args; /* will hold data from mount request */
1996-02-09 22:13:39 +03:00
/* msdosfs specific mount control block */
struct msdosfsmount *pmp = NULL;
1995-03-09 15:05:21 +03:00
size_t size;
int error, flags;
1994-12-14 19:33:18 +03:00
mode_t accessmode;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
1996-02-09 22:13:39 +03:00
error = copyin(data, (caddr_t)&args, sizeof(struct msdosfs_args));
if (error)
return (error);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
1994-12-14 19:33:18 +03:00
* If updating, check whether changing from read-only to
* read/write; if there is no device name, that's all we do.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
if (mp->mnt_flag & MNT_UPDATE) {
1994-09-22 06:34:49 +04:00
pmp = VFSTOMSDOSFS(mp);
1994-07-17 01:32:06 +04:00
error = 0;
1995-10-15 18:34:19 +03:00
if (!(pmp->pm_flags & MSDOSFSMNT_RONLY) && (mp->mnt_flag & MNT_RDONLY)) {
1994-07-17 01:32:06 +04:00
flags = WRITECLOSE;
if (mp->mnt_flag & MNT_FORCE)
flags |= FORCECLOSE;
error = vflush(mp, NULLVP, flags);
}
if (!error && (mp->mnt_flag & MNT_RELOAD))
/* not yet implemented */
error = EOPNOTSUPP;
1994-07-17 01:32:06 +04:00
if (error)
return (error);
1995-10-15 18:34:19 +03:00
if ((pmp->pm_flags & MSDOSFSMNT_RONLY) && (mp->mnt_flag & MNT_WANTRDWR)) {
1994-12-14 19:33:18 +03:00
/*
* If upgrade to read-write by non-root, then verify
* that user has necessary permissions on the device.
*/
if (p->p_ucred->cr_uid != 0) {
devvp = pmp->pm_devvp;
1998-03-01 05:20:01 +03:00
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
1996-02-09 22:13:39 +03:00
error = VOP_ACCESS(devvp, VREAD | VWRITE,
p->p_ucred, p);
1998-03-01 05:20:01 +03:00
VOP_UNLOCK(devvp, 0);
if (error)
1994-12-14 19:33:18 +03:00
return (error);
}
1995-10-15 18:34:19 +03:00
pmp->pm_flags &= ~MSDOSFSMNT_RONLY;
1994-12-14 19:33:18 +03:00
}
1994-07-17 01:32:06 +04:00
if (args.fspec == 0) {
1995-10-15 18:34:19 +03:00
#ifdef __notyet__ /* doesn't work correctly with current mountd XXX */
if (args.flags & MSDOSFSMNT_MNTOPT) {
pmp->pm_flags &= ~MSDOSFSMNT_MNTOPT;
pmp->pm_flags |= args.flags & MSDOSFSMNT_MNTOPT;
if (pmp->pm_flags & MSDOSFSMNT_NOWIN95)
pmp->pm_flags |= MSDOSFSMNT_SHORTNAME;
}
#endif
1994-07-17 01:32:06 +04:00
/*
* Process export requests.
*/
return (vfs_export(mp, &pmp->pm_export, &args.export));
1994-07-17 01:32:06 +04:00
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
/*
* Not an update, or updating the name: look up the name
1994-12-14 19:33:18 +03:00
* and verify that it refers to a sensible block device.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
1994-07-17 01:32:06 +04:00
NDINIT(ndp, LOOKUP, FOLLOW, UIO_USERSPACE, args.fspec, p);
1996-02-09 22:13:39 +03:00
if ((error = namei(ndp)) != 0)
return (error);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
devvp = ndp->ni_vp;
1994-12-14 19:33:18 +03:00
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
if (devvp->v_type != VBLK) {
1994-07-17 01:32:06 +04:00
vrele(devvp);
return (ENOTBLK);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
if (major(devvp->v_rdev) >= nblkdev) {
vrele(devvp);
return (ENXIO);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
/*
1994-12-14 19:33:18 +03:00
* If mount by non-root, then verify that user has necessary
* permissions on the device.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
1994-12-14 19:33:18 +03:00
if (p->p_ucred->cr_uid != 0) {
accessmode = VREAD;
if ((mp->mnt_flag & MNT_RDONLY) == 0)
accessmode |= VWRITE;
1998-03-01 05:20:01 +03:00
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
1996-02-09 22:13:39 +03:00
error = VOP_ACCESS(devvp, accessmode, p->p_ucred, p);
1998-03-01 05:20:01 +03:00
VOP_UNLOCK(devvp, 0);
1996-02-09 22:13:39 +03:00
if (error) {
1998-03-01 05:20:01 +03:00
vrele(devvp);
1994-12-14 19:33:18 +03:00
return (error);
}
}
if ((mp->mnt_flag & MNT_UPDATE) == 0) {
error = msdosfs_mountfs(devvp, mp, p, &args);
#ifdef MSDOSFS_DEBUG /* only needed for the printf below */
pmp = VFSTOMSDOSFS(mp);
#endif
} else {
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
if (devvp != pmp->pm_devvp)
1994-12-14 19:33:18 +03:00
error = EINVAL; /* needs translation */
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
else
vrele(devvp);
1994-12-14 19:33:18 +03:00
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
if (error) {
vrele(devvp);
return (error);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
if ((error = update_mp(mp, &args)) != 0) {
msdosfs_unmount(mp, MNT_FORCE, p);
return error;
1995-10-15 18:34:19 +03:00
}
(void) copyinstr(path, mp->mnt_stat.f_mntonname, MNAMELEN - 1, &size);
bzero(mp->mnt_stat.f_mntonname + size, MNAMELEN - size);
(void) copyinstr(args.fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1,
&size);
bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size);
#ifdef MSDOSFS_DEBUG
printf("msdosfs_mount(): mp %p, pmp %p, inusemap %p\n", mp, pmp, pmp->pm_inusemap);
#endif
return (0);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
msdosfs_mountfs(devvp, mp, p, argp)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
struct vnode *devvp;
struct mount *mp;
struct proc *p;
struct msdosfs_args *argp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
{
struct msdosfsmount *pmp;
struct buf *bp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
dev_t dev = devvp->v_rdev;
struct partinfo dpart;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
union bootsector *bsp;
struct byte_bpb33 *b33;
struct byte_bpb50 *b50;
struct byte_bpb710 *b710;
extern struct vnode *rootvp;
1995-10-15 18:34:19 +03:00
u_int8_t SecPerClust;
int ronly, error;
1996-02-09 22:13:39 +03:00
int bsize = 0, dtype = 0, tmp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* Disallow multiple mounts of the same device.
* Disallow mounting of a device that is currently in use
* (except for root, which might share swap device for miniroot).
* Flush out any old buffers remaining from a previous use.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
1996-02-09 22:13:39 +03:00
if ((error = vfs_mountedon(devvp)) != 0)
return (error);
if (vcount(devvp) > 1 && devvp != rootvp)
return (EBUSY);
1996-02-09 22:13:39 +03:00
if ((error = vinvalbuf(devvp, V_SAVE, p->p_ucred, p, 0, 0)) != 0)
return (error);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
ronly = (mp->mnt_flag & MNT_RDONLY) != 0;
1996-02-09 22:13:39 +03:00
error = VOP_OPEN(devvp, ronly ? FREAD : FREAD|FWRITE, FSCRED, p);
if (error)
return (error);
bp = NULL; /* both used in error_exit */
pmp = NULL;
if (argp->flags & MSDOSFSMNT_GEMDOSFS) {
/*
* We need the disklabel to calculate the size of a FAT entry
* later on. Also make sure the partition contains a filesystem
* of type FS_MSDOS. This doesn't work for floppies, so we have
* to check for them too.
*
* At least some parts of the msdos fs driver seem to assume
* that the size of a disk block will always be 512 bytes.
* Let's check it...
*/
1996-02-09 22:13:39 +03:00
error = VOP_IOCTL(devvp, DIOCGPART, (caddr_t)&dpart,
FREAD, NOCRED, p);
if (error)
goto error_exit;
tmp = dpart.part->p_fstype;
dtype = dpart.disklab->d_type;
bsize = dpart.disklab->d_secsize;
if (bsize != 512 || (dtype!=DTYPE_FLOPPY && tmp!=FS_MSDOS)) {
error = EINVAL;
goto error_exit;
}
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* Read the boot sector of the filesystem, and then check the
* boot signature. If not a dos boot sector then error out.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
1996-02-09 22:13:39 +03:00
if ((error = bread(devvp, 0, 512, NOCRED, &bp)) != 0)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
goto error_exit;
bp->b_flags |= B_AGE;
bsp = (union bootsector *)bp->b_data;
b33 = (struct byte_bpb33 *)bsp->bs33.bsBPB;
b50 = (struct byte_bpb50 *)bsp->bs50.bsBPB;
b710 = (struct byte_bpb710 *)bsp->bs710.bsPBP;
if (!(argp->flags & MSDOSFSMNT_GEMDOSFS)) {
if (bsp->bs50.bsBootSectSig0 != BOOTSIG0
|| bsp->bs50.bsBootSectSig1 != BOOTSIG1) {
error = EINVAL;
goto error_exit;
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
pmp = malloc(sizeof *pmp, M_MSDOSFSMNT, M_WAITOK);
1994-07-17 01:32:06 +04:00
bzero((caddr_t)pmp, sizeof *pmp);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_mountp = mp;
/*
* Compute several useful quantities from the bpb in the
* bootsector. Copy in the dos 5 variant of the bpb then fix up
* the fields that are different between dos 5 and dos 3.3.
*/
1995-10-15 18:34:19 +03:00
SecPerClust = b50->bpbSecPerClust;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_BytesPerSec = getushort(b50->bpbBytesPerSec);
pmp->pm_ResSectors = getushort(b50->bpbResSectors);
pmp->pm_FATs = b50->bpbFATs;
pmp->pm_RootDirEnts = getushort(b50->bpbRootDirEnts);
pmp->pm_Sectors = getushort(b50->bpbSectors);
pmp->pm_FATsecs = getushort(b50->bpbFATsecs);
pmp->pm_SecPerTrack = getushort(b50->bpbSecPerTrack);
pmp->pm_Heads = getushort(b50->bpbHeads);
pmp->pm_Media = b50->bpbMedia;
if (!(argp->flags & MSDOSFSMNT_GEMDOSFS)) {
/* XXX - We should probably check more values here */
if (!pmp->pm_BytesPerSec || !SecPerClust
|| pmp->pm_Heads > 255 || pmp->pm_SecPerTrack > 63) {
error = EINVAL;
goto error_exit;
}
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
if (pmp->pm_Sectors == 0) {
pmp->pm_HiddenSects = getulong(b50->bpbHiddenSecs);
pmp->pm_HugeSectors = getulong(b50->bpbHugeSectors);
1994-07-17 01:32:06 +04:00
} else {
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_HiddenSects = getushort(b33->bpbHiddenSecs);
pmp->pm_HugeSectors = pmp->pm_Sectors;
}
if (pmp->pm_HugeSectors > 0xffffffff / pmp->pm_BytesPerSec + 1) {
/*
* We cannot deal currently with this size of disk
* due to fileid limitations (see msdosfs_getattr and
* msdosfs_readdir)
*/
error = EINVAL;
goto error_exit;
}
if (pmp->pm_RootDirEnts == 0) {
if (bsp->bs710.bsBootSectSig2 != BOOTSIG2
|| bsp->bs710.bsBootSectSig3 != BOOTSIG3
|| pmp->pm_Sectors
|| pmp->pm_FATsecs
|| getushort(b710->bpbFSVers)) {
error = EINVAL;
goto error_exit;
}
pmp->pm_fatmask = FAT32_MASK;
pmp->pm_fatmult = 4;
pmp->pm_fatdiv = 1;
pmp->pm_FATsecs = getulong(b710->bpbBigFATsecs);
if (getushort(b710->bpbExtFlags) & FATMIRROR)
pmp->pm_curfat = getushort(b710->bpbExtFlags) & FATNUM;
else
pmp->pm_flags |= MSDOSFS_FATMIRROR;
} else
pmp->pm_flags |= MSDOSFS_FATMIRROR;
if (argp->flags & MSDOSFSMNT_GEMDOSFS) {
if (FAT32(pmp)) {
/*
* GEMDOS doesn't know fat32.
*/
error = EINVAL;
goto error_exit;
}
/*
* Check a few values (could do some more):
* - logical sector size: power of 2, >= block size
* - sectors per cluster: power of 2, >= 1
* - number of sectors: >= 1, <= size of partition
*/
if ( (SecPerClust == 0)
|| (SecPerClust & (SecPerClust - 1))
|| (pmp->pm_BytesPerSec < bsize)
|| (pmp->pm_BytesPerSec & (pmp->pm_BytesPerSec - 1))
|| (pmp->pm_HugeSectors == 0)
|| (pmp->pm_HugeSectors * (pmp->pm_BytesPerSec / bsize)
> dpart.part->p_size)
) {
error = EINVAL;
goto error_exit;
}
/*
* XXX - Many parts of the msdos fs driver seem to assume that
* the number of bytes per logical sector (BytesPerSec) will
* always be the same as the number of bytes per disk block
* Let's pretend it is.
*/
tmp = pmp->pm_BytesPerSec / bsize;
pmp->pm_BytesPerSec = bsize;
pmp->pm_HugeSectors *= tmp;
pmp->pm_HiddenSects *= tmp;
pmp->pm_ResSectors *= tmp;
pmp->pm_Sectors *= tmp;
pmp->pm_FATsecs *= tmp;
SecPerClust *= tmp;
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_fatblk = pmp->pm_ResSectors;
if (FAT32(pmp)) {
pmp->pm_rootdirblk = getulong(b710->bpbRootClust);
pmp->pm_firstcluster = pmp->pm_fatblk
+ (pmp->pm_FATs * pmp->pm_FATsecs);
pmp->pm_fsinfo = getushort(b710->bpbFSInfo);
} else {
pmp->pm_rootdirblk = pmp->pm_fatblk +
(pmp->pm_FATs * pmp->pm_FATsecs);
pmp->pm_rootdirsize = (pmp->pm_RootDirEnts * sizeof(struct direntry)
+ pmp->pm_BytesPerSec - 1)
/ pmp->pm_BytesPerSec;/* in sectors */
pmp->pm_firstcluster = pmp->pm_rootdirblk + pmp->pm_rootdirsize;
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_nmbrofclusters = (pmp->pm_HugeSectors - pmp->pm_firstcluster) /
1995-10-15 18:34:19 +03:00
SecPerClust;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_maxcluster = pmp->pm_nmbrofclusters + 1;
pmp->pm_fatsize = pmp->pm_FATsecs * pmp->pm_BytesPerSec;
if (argp->flags & MSDOSFSMNT_GEMDOSFS) {
if ((pmp->pm_nmbrofclusters <= (0xff0 - 2))
&& ((dtype == DTYPE_FLOPPY) || ((dtype == DTYPE_VNODE)
&& ((pmp->pm_Heads == 1) || (pmp->pm_Heads == 2))))
) {
pmp->pm_fatmask = FAT12_MASK;
pmp->pm_fatmult = 3;
pmp->pm_fatdiv = 2;
} else {
pmp->pm_fatmask = FAT16_MASK;
pmp->pm_fatmult = 2;
pmp->pm_fatdiv = 1;
}
} else if (pmp->pm_fatmask == 0) {
if (pmp->pm_maxcluster
<= ((CLUST_RSRVD - CLUST_FIRST) & FAT12_MASK)) {
/*
* This will usually be a floppy disk. This size makes
* sure that one fat entry will not be split across
* multiple blocks.
*/
pmp->pm_fatmask = FAT12_MASK;
pmp->pm_fatmult = 3;
pmp->pm_fatdiv = 2;
} else {
pmp->pm_fatmask = FAT16_MASK;
pmp->pm_fatmult = 2;
pmp->pm_fatdiv = 1;
}
}
if (FAT12(pmp))
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_fatblocksize = 3 * pmp->pm_BytesPerSec;
else
pmp->pm_fatblocksize = MAXBSIZE;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_fatblocksec = pmp->pm_fatblocksize / pmp->pm_BytesPerSec;
1995-10-15 18:34:19 +03:00
pmp->pm_bnshift = ffs(pmp->pm_BytesPerSec) - 1;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
1995-10-15 18:34:19 +03:00
* Compute mask and shift value for isolating cluster relative byte
* offsets and cluster numbers from a file offset.
*/
1995-10-15 18:34:19 +03:00
pmp->pm_bpcluster = SecPerClust * pmp->pm_BytesPerSec;
pmp->pm_crbomask = pmp->pm_bpcluster - 1;
pmp->pm_cnshift = ffs(pmp->pm_bpcluster) - 1;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
1995-10-15 18:34:19 +03:00
* Check for valid cluster size
* must be a power of 2
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
1995-10-15 18:34:19 +03:00
if (pmp->pm_bpcluster ^ (1 << pmp->pm_cnshift)) {
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
error = EINVAL;
goto error_exit;
}
/*
* Release the bootsector buffer.
*/
brelse(bp);
bp = NULL;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* Check FSInfo.
*/
if (pmp->pm_fsinfo) {
struct fsinfo *fp;
if ((error = bread(devvp, pmp->pm_fsinfo, 1024, NOCRED, &bp)) != 0)
goto error_exit;
fp = (struct fsinfo *)bp->b_data;
if (!bcmp(fp->fsisig1, "RRaA", 4)
&& !bcmp(fp->fsisig2, "rrAa", 4)
&& !bcmp(fp->fsisig3, "\0\0\125\252", 4)
&& !bcmp(fp->fsisig4, "\0\0\125\252", 4))
pmp->pm_nxtfree = getulong(fp->fsinxtfree);
else
pmp->pm_fsinfo = 0;
brelse(bp);
bp = NULL;
}
/*
* Check and validate (or perhaps invalidate?) the fsinfo structure? XXX
*/
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* Allocate memory for the bitmap of allocated clusters, and then
* fill it in.
*/
pmp->pm_inusemap = malloc(((pmp->pm_maxcluster + N_INUSEBITS - 1)
/ N_INUSEBITS)
* sizeof(*pmp->pm_inusemap),
M_MSDOSFSFAT, M_WAITOK);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* fillinusemap() needs pm_devvp.
*/
pmp->pm_dev = dev;
pmp->pm_devvp = devvp;
/*
* Have the inuse map filled in.
*/
1996-02-09 22:13:39 +03:00
if ((error = fillinusemap(pmp)) != 0)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
goto error_exit;
/*
* If they want fat updates to be synchronous then let them suffer
* the performance degradation in exchange for the on disk copy of
* the fat being correct just about all the time. I suppose this
* would be a good thing to turn on if the kernel is still flakey.
*/
1995-10-15 18:34:19 +03:00
if (mp->mnt_flag & MNT_SYNCHRONOUS)
pmp->pm_flags |= MSDOSFSMNT_WAITONFAT;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* Finish up.
*/
1995-10-15 18:34:19 +03:00
if (ronly)
pmp->pm_flags |= MSDOSFSMNT_RONLY;
else
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_fmod = 1;
mp->mnt_data = (qaddr_t)pmp;
1994-04-14 08:05:45 +04:00
mp->mnt_stat.f_fsid.val[0] = (long)dev;
mp->mnt_stat.f_fsid.val[1] = makefstype(MOUNT_MSDOS);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
mp->mnt_flag |= MNT_LOCAL;
#ifdef QUOTA
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* If we ever do quotas for DOS filesystems this would be a place
* to fill in the info in the msdosfsmount structure. You dolt,
* quotas on dos filesystems make no sense because files have no
* owners on dos filesystems. of course there is some empty space
* in the directory entry where we could put uid's and gid's.
*/
#endif
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
devvp->v_specflags |= SI_MOUNTEDON;
return (0);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
error_exit:;
if (bp)
brelse(bp);
(void) VOP_CLOSE(devvp, ronly ? FREAD : FREAD|FWRITE, NOCRED, p);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
if (pmp) {
if (pmp->pm_inusemap)
free(pmp->pm_inusemap, M_MSDOSFSFAT);
free(pmp, M_MSDOSFSMNT);
mp->mnt_data = (qaddr_t)0;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
return (error);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
msdosfs_start(mp, flags, p)
struct mount *mp;
int flags;
struct proc *p;
{
return (0);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
/*
* Unmount the filesystem described by mp.
*/
int
msdosfs_unmount(mp, mntflags, p)
struct mount *mp;
int mntflags;
struct proc *p;
{
1994-12-14 19:33:18 +03:00
struct msdosfsmount *pmp;
int error, flags;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
flags = 0;
if (mntflags & MNT_FORCE)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
flags |= FORCECLOSE;
#ifdef QUOTA
#endif
1996-02-09 22:13:39 +03:00
if ((error = vflush(mp, NULLVP, flags)) != 0)
return (error);
1994-12-14 19:33:18 +03:00
pmp = VFSTOMSDOSFS(mp);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp->pm_devvp->v_specflags &= ~SI_MOUNTEDON;
#ifdef MSDOSFS_DEBUG
{
struct vnode *vp = pmp->pm_devvp;
printf("msdosfs_umount(): just before calling VOP_CLOSE()\n");
printf("flag %08lx, usecount %d, writecount %d, holdcnt %ld\n",
vp->v_flag, vp->v_usecount, vp->v_writecount, vp->v_holdcnt);
printf("lastr %d, id %lu, mount %p, op %p\n",
vp->v_lastr, vp->v_id, vp->v_mount, vp->v_op);
printf("freef %p, freeb %p, mount %p\n",
vp->v_freelist.tqe_next, vp->v_freelist.tqe_prev,
vp->v_mount);
printf("cleanblkhd %p, dirtyblkhd %p, numoutput %ld, type %d\n",
vp->v_cleanblkhd.lh_first,
vp->v_dirtyblkhd.lh_first,
vp->v_numoutput, vp->v_type);
printf("union %p, tag %d, data[0] %08x, data[1] %08x\n",
vp->v_socket, vp->v_tag,
((u_int *)vp->v_data)[0],
((u_int *)vp->v_data)[1]);
}
#endif
1995-10-30 22:06:15 +03:00
error = VOP_CLOSE(pmp->pm_devvp,
pmp->pm_flags & MSDOSFSMNT_RONLY ? FREAD : FREAD|FWRITE, NOCRED, p);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
vrele(pmp->pm_devvp);
free(pmp->pm_inusemap, M_MSDOSFSFAT);
free(pmp, M_MSDOSFSMNT);
mp->mnt_data = (qaddr_t)0;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
mp->mnt_flag &= ~MNT_LOCAL;
return (error);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
msdosfs_root(mp, vpp)
struct mount *mp;
struct vnode **vpp;
{
1994-09-22 06:34:49 +04:00
struct msdosfsmount *pmp = VFSTOMSDOSFS(mp);
struct denode *ndep;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
int error;
#ifdef MSDOSFS_DEBUG
printf("msdosfs_root(); mp %p, pmp %p\n", mp, pmp);
#endif
if ((error = deget(pmp, MSDOSFSROOT, MSDOSFSROOT_OFS, &ndep)) != 0)
return (error);
*vpp = DETOV(ndep);
return (0);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
msdosfs_quotactl(mp, cmds, uid, arg, p)
struct mount *mp;
int cmds;
uid_t uid;
caddr_t arg;
struct proc *p;
{
#ifdef QUOTA
return (EOPNOTSUPP);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
#else
return (EOPNOTSUPP);
#endif
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
msdosfs_statfs(mp, sbp, p)
struct mount *mp;
struct statfs *sbp;
struct proc *p;
{
struct msdosfsmount *pmp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
pmp = VFSTOMSDOSFS(mp);
1994-04-14 08:05:45 +04:00
#ifdef COMPAT_09
sbp->f_type = 4;
#else
sbp->f_type = 0;
#endif
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
sbp->f_bsize = pmp->pm_bpcluster;
sbp->f_iosize = pmp->pm_bpcluster;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
sbp->f_blocks = pmp->pm_nmbrofclusters;
sbp->f_bfree = pmp->pm_freeclustercount;
sbp->f_bavail = pmp->pm_freeclustercount;
1994-07-17 01:32:06 +04:00
sbp->f_files = pmp->pm_RootDirEnts; /* XXX */
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
sbp->f_ffree = 0; /* what to put in here? */
if (sbp != &mp->mnt_stat) {
bcopy(mp->mnt_stat.f_mntonname, sbp->f_mntonname, MNAMELEN);
bcopy(mp->mnt_stat.f_mntfromname, sbp->f_mntfromname, MNAMELEN);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
strncpy(sbp->f_fstypename, mp->mnt_op->vfs_name, MFSNAMELEN);
return (0);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
1994-07-17 01:32:06 +04:00
msdosfs_sync(mp, waitfor, cred, p)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
struct mount *mp;
int waitfor;
1994-07-17 01:32:06 +04:00
struct ucred *cred;
struct proc *p;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
{
1998-03-01 05:20:01 +03:00
struct vnode *vp, *nvp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
struct denode *dep;
1994-09-19 23:28:07 +04:00
struct msdosfsmount *pmp = VFSTOMSDOSFS(mp);
int error, allerror = 0;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
* If we ever switch to not updating all of the fats all the time,
* this would be the place to update them from the first one.
*/
if (pmp->pm_fmod != 0)
1995-10-15 18:34:19 +03:00
if (pmp->pm_flags & MSDOSFSMNT_RONLY)
panic("msdosfs_sync: rofs mod");
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
else {
/* update fats here */
}
/*
1994-09-19 23:28:07 +04:00
* Write back each (modified) denode.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
1998-03-01 05:20:01 +03:00
simple_lock(&mntvnode_slock);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
loop:
1998-03-01 05:20:01 +03:00
for (vp = mp->mnt_vnodelist.lh_first; vp != NULL; vp = nvp) {
1994-09-19 23:28:07 +04:00
/*
* If the vnode that we are about to sync is no longer
* assoicated with this mount point, start over.
*/
if (vp->v_mount != mp)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
goto loop;
1998-03-01 05:20:01 +03:00
simple_lock(&vp->v_interlock);
nvp = vp->v_mntvnodes.le_next;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
dep = VTODE(vp);
if (((dep->de_flag
& (DE_ACCESS | DE_CREATE | DE_UPDATE | DE_MODIFIED)) == 0)
1998-03-01 05:20:01 +03:00
&& (vp->v_dirtyblkhd.lh_first == NULL)) {
simple_unlock(&vp->v_interlock);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
continue;
1998-03-01 05:20:01 +03:00
}
simple_unlock(&mntvnode_slock);
error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK);
if (error) {
simple_lock(&mntvnode_slock);
if (error == ENOENT)
goto loop;
continue;
}
1996-02-09 22:13:39 +03:00
if ((error = VOP_FSYNC(vp, cred, waitfor, p)) != 0)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
allerror = error;
1994-09-19 23:28:07 +04:00
vput(vp);
1998-03-01 05:20:01 +03:00
simple_lock(&mntvnode_slock);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
1998-03-01 05:20:01 +03:00
simple_unlock(&mntvnode_slock);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
/*
1994-09-19 23:28:07 +04:00
* Force stale file system control information to be flushed.
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*/
1996-02-09 22:13:39 +03:00
if ((error = VOP_FSYNC(pmp->pm_devvp, cred, waitfor, p)) != 0)
1994-09-19 23:28:07 +04:00
allerror = error;
#ifdef QUOTA
#endif
return (allerror);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
1994-07-17 01:32:06 +04:00
msdosfs_fhtovp(mp, fhp, nam, vpp, exflagsp, credanonp)
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
struct mount *mp;
struct fid *fhp;
1994-07-17 01:32:06 +04:00
struct mbuf *nam;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
struct vnode **vpp;
1994-07-17 01:32:06 +04:00
int *exflagsp;
struct ucred **credanonp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
{
1994-09-22 06:34:49 +04:00
struct msdosfsmount *pmp = VFSTOMSDOSFS(mp);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
struct defid *defhp = (struct defid *) fhp;
struct denode *dep;
1994-07-17 01:32:06 +04:00
struct netcred *np;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
int error;
1994-07-17 01:32:06 +04:00
np = vfs_export_lookup(mp, &pmp->pm_export, nam);
if (np == NULL)
return (EACCES);
error = deget(pmp, defhp->defid_dirclust, defhp->defid_dirofs, &dep);
1994-07-17 01:32:06 +04:00
if (error) {
*vpp = NULLVP;
return (error);
1994-07-17 01:32:06 +04:00
}
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
*vpp = DETOV(dep);
1994-07-17 01:32:06 +04:00
*exflagsp = np->netc_exflags;
*credanonp = &np->netc_anon;
return (0);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
int
msdosfs_vptofh(vp, fhp)
struct vnode *vp;
struct fid *fhp;
{
struct denode *dep;
struct defid *defhp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
dep = VTODE(vp);
defhp = (struct defid *)fhp;
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
defhp->defid_len = sizeof(struct defid);
defhp->defid_dirclust = dep->de_dirclust;
defhp->defid_dirofs = dep->de_diroffset;
/* defhp->defid_gen = dep->de_gen; */
return (0);
1994-07-17 01:32:06 +04:00
}
int
msdosfs_vget(mp, ino, vpp)
struct mount *mp;
ino_t ino;
struct vnode **vpp;
{
return (EOPNOTSUPP);
brought in fixed/renamed/matching MS-DOS FS code, from Jeff Polk <polk@bsdi.com>. His notes are as follows: ------------------------------------------------------------------------------ July 22, 1993 - Changed name of entire package from PCFS to MSDOSFS - Fixed bugs: root directory size in clusters instead of bytes growing directory didn't update in-core size link, symlink, mknod didn't free locked parent (deadlock) lookup returned real error on create and rename instead of EJUSTRETURN rename changed `.' entry in child instead of name entry in parent rename removed `.' entry in child instead of removing entry in parent when moving a directory from one dir to another createde() left new node locked when write of parent failed (deadlock) removede() decremented refcount even on error (rmdir's which failed due to write errors left in-core cache entries inconsistent) changed validation for filesystem to not check for the boot signature since some disks (e.g., mtools) aren't bootable directories are always show current time as modify time (needed for NFS export since DOS never updates dir mod times -- ctime is true create time). - Added support for cookies changes to the readdir() vnode interface (#ifdef __bsdi__) - Punted on the whole problem of inode generation numbers. This means that there's a chance of using a stale file handle to access a new file, but it doesn't appear to be the common case, and I don't see how to generate reasonable generation numbers without changing something on the disk (which is the way the SVR4 filesystem survival kit guys did it). I don't think it would be very safe to change the on-disk format. Jeff Polk (polk@BSDI.COM) ------------------------------------------------------------------------------
1993-08-13 15:35:13 +04:00
}
1998-03-01 05:20:01 +03:00
int
msdosfs_sysctl(name, namelen, oldp, oldlenp, newp, newlen, p)
int *name;
u_int namelen;
void *oldp;
size_t *oldlenp;
void *newp;
size_t newlen;
struct proc *p;
{
return (EOPNOTSUPP);
}