DOC HOME SITE MAP MAN PAGES GNU INFO SEARCH PRINT BOOK
 

mount(ADM)


mount, umount -- mount and unmount filesystems and remote resources

Syntax

/etc/mount [-m] [-r] [-v] [-l] [-w] [-o options] [-f fstyp[,options]] special|resource directory

/etc/umount [-m] special|directory|resource

Description

mount announces to the system that a removable filesystem or resource is present on the special device. (Filesystems other than root (/) are considered removable in the sense that they can be either available to users or unavailable.)

A block special device or a remote resource can be mounted at the point directory in an existing filesystem. directory becomes the name of the root of the newly mounted filesystem or resource. directory must already exist and should be empty. If directory contains files, they will appear to have been removed after the mount; they will reappear after an unmount using umount.

A unique resource can be mounted only once (multiple mounts are not allowed).

mount and umount use a table of mounted devices /etc/mnttab to add, query, and remove entries.

mount invoked without any arguments displays:

mount accepts the following options:


-f fstyp[,options]
indicates the type of filesystem (fstyp) to be mounted. Filesystem types are listed in the ``Filesystem types'' section of this manual page.

If you omit this option, mount uses the type defined for the filesystem in its /etc/default/filesys entry (see filesys(F)). If the filesystem being mounted does not have an entry in filesys, mount calls statfs(S) to read the filesystem type from its superblock.

You can specify filesystem-specific options after the filesystem type instead of using the -o option. The recommended method is to use -o.

Separate the modifiers from each other and from the filesystem type using commas ``,''; do not include spaces (see ``Examples'').


-m
Instructs mount or umount to refrain from examining or updating the mount table (/etc/mnttab). This option is generally not recommended for casual use.

-l
With no other options or arguments, only displays local (not network-mounted) filesystems.

-w
Mounts the filesystem for read/write access. Overrides settings specified in the /etc/default/filesys file.

-o options
specifies generic and filesystem-specific mount options. The following generic options are available:

exec
allow all executable files to be executed (default).

noexec
do not allow files to be executed.

remount
change the generic options on an already-mounted filesystem without having to unmount it.

You can specify any other generic option with remount. You must specify any non-default generic option (noexec, ro, nosuid, or trunc) explicitly if it is to take effect. Otherwise, the filesystem reverts to the corresponding default option instead.

Note that it is not possible to change a filesystem from a read/write state (rw) to a read-only state (ro).


rw
mount for reading and writing (default).

ro
mount for read-only. This is equivalent to specifying the -r option.

suid
obey setuid and setgid bits when executing a file using the exec(S) system call (default).

nosuid
ignore setuid and setgid bits when executing a file using the exec system call. The suid and nosuid options only apply to Rock Ridge (CD-ROM) and NFS filesystems.

maxvdepth=n
the maximum number of previous versions, n, which will be maintained for versioned files in a filesystem. n must be a positive integer. A value of 0 disables versioning. A value of ``unlimited'' enables the maximum number of versions possible for the filesystem. Setting this value to 1 or higher effectively enables versioning for the filesystem. If this value is not set, default values are specified in the kernel tunable parameter MAXVDEPTH. This is supported only on HTFS(TM) and DTFS(TM) filesystem types.

minvtime=n
the minimum time (in seconds) that a file must exist before it can be versioned. If n is 0, there is no minimum time. If this value is not set, default values are specified in the kernel tunable parameter MINVTIME.This is supported only on HTFS and DTFS filesystem types.

trunc
truncate long filenames silently. This disables the path variable _POSIX_NO_TRUNC (see the description in getconf(C)). Do not use this option if you require FIPS or XPG4-conformance.

notrunc
display the error message Filename too long if a program attempts to access filenames which are longer than the filesystem allows (default).

Any system call which attempts to access or create filenames that are too long will fail and set ENAMETOOLONG in errno(S). This usually causes the program to display the above error message and exit.

Multiple options are specified as a comma-separated list. When options conflict, the last option takes precedence.

An option only has an effect if it is valid for the filesystem type. For example, a DOS filesystem understands options such as ro, remount, and exec but it does not understand options such as trunc and suid.

See the description of the -f option for the descriptions of valid filesystem-specific option modifiers that can be used with -o.


-r
indicates that a special device or resource is to be mounted for read-only. Physically write-protected filesystems, such as floppy disks with write-protect tabs, and read-only advertised filesystems must be mounted in this way. Otherwise errors occur when access times are updated, whether or not any explicit write is attempted.

The -o ro option is equivalent to -r.


-v
displays mount information verbosely.

Filesystem types

Filesystem types that can be mounted are:

AFS
Acer Fast Filesystems.

The following option modifiers are valid with AFS, S51K, EAFS and HTFS filesystems:


nochkpt
Mount the filesystem with checkpointing disabled. Checkpointing can speed reboot time after an unexpected system error by reducing the chances that the filesystem will need to be checked with fsck(ADM). Disabling checkpointing can improve overall system performance.

nolog
Mount the filesystem with logging disabled. Logging is used to reduce the amount of time needed to check a filesystem if a system crash occurs. The logging mechanism keeps track of all modified data; this way, only affected areas of the disk need to be checked. The S51K filesystem does not support transaction logging.

tmp
Mount the filesystem as a temporary filesystem. Temporary filesystems are updated less frequently, and are recommended for use on filesystems containing temporary data only. If used on /tmp, the overall system performance may be improved. If this option is specified, then checkpointing will be disabled automatically.

DOS
DOS filesystems.

The following option modifier is available:


lower
Display filenames in lowercase. The default is to use uppercase.

DTFS
Desktop Filesystems.

The following option modifiers are available:


nocomp
Do not compress data written into this filesystem.

sync
Force data to be written synchronously to disk whenever a file is closed.

EAFS
Extended Acer Fast Filesystems.

The nochkpt, nolog and tmp option modifiers, described for the AFS filesystem type, are also valid for EAFS filesystems.


HS
High Sierra filesystems.

With the exception of uid, gid, and dev, each option modifier should be specified no more than once (later invocations of a modifier will override earlier ones). If more than one mapping is specified for a file or a value, the first instance is used.

The following option modifiers are valid with High Sierra and ISO 9660 filesystems:


hide
Do not show hidden files.

lower
Display filenames in lowercase and suppress trailing ``.''s. The default is to use the filename as recorded on the CD-ROM. If filenames have been recorded as POSIX(TM) -style, this modifier is ineffective.

version
Display file version numbers. The default is not to display these. If filenames have been recorded as POSIX-style, this modifier is ineffective.

session=n
On multisession discs, the most recent session is mounted by default. At this time the driver only supports mounting the first and last sessions. You can access the last (most recent) session explicitly, or by using the value of 0. For example, on a disc with five sessions, a value of 1 indicates the first session, while 0 or 5 apply to the last session.

sector=n
Allows the starting sector to be specified (which can be useful when the start and end sectors are known, or when attempting to mount a disc that has been written in a non-standard way).

In addition, the following option modifier is valid for High Sierra filesystems:


relax
Apply the checks that were previously applied in the SCO Open Desktop Development System. Use this option to mount CD-ROMs that are not completely conformant to ISO 9660.

In the following option modifiers, the term XAR (extended attribute record) refers to the optional data structure used within an ISO 9660 format CD-ROM for recording file attributes:


defuid=UID|user

defgid=GID|group
defuid defines the user ID for files and directories recorded without a final unrestricted XAR (default is the user ID of the mount point).

defgid is similar to defuid, but sets the default group ID.

If specified, user (group) must be a valid user (group) name; UID (GID) must be a valid user (group) ID in the range 0 to 60000.


filemode=mode

dirmode=mode
filemode sets the permissions for files recorded without a final unrestricted XAR.

dirmode sets the permissions for directories recorded without a final unrestricted XAR.

mode is an octal value produced by ORing the following values (default is 0555):

Value Permission set
0400 read by owner
0100 execute by owner
0040 read by group
0010 execute by group
0004 read by others
0001 execute by others

 Value                  Permission set
 0400                   read by owner
 0100                   execute by owner
 0040                   read by group
 0010                   execute by group
 0004                   read by others
 0001                   execute by others

search
Grant search permissions for owner, group, and others on all directories recorded with an XAR regardless of the permissions recorded in it. Directories without an XAR are controlled by the dirmode option.

HTFS
High Throughput Filesystems.

The nochkpt, nolog and tmp option modifiers, described for the AFS filesystem type, are also valid with HTFS filesystems.


ISO9660
ISO 9660 CD-ROM filesystems.

With the exception of relax, the option modifiers described for the HS filesystem are also valid with ISO 9660 filesystems.


NFS
Networked Filesystems.

Specify the remote resource in the form hostname:pathname. The following option modifiers are available:


soft
Return error if the server does not respond.

bg
Background this mount. (This is recommended for automatic mounts done during system startup.)

rsize=n
Set the read buffer size to n bytes. The default size is 8192 bytes.

wsize=n
Set the write buffer size to n bytes. The default size is 8192 bytes.

timeo=n
Set the initial NFS(TM) timeout to n tenths of a second. The default is 300 tenths of a second.

retrans=n
Set the number of NFS retransmissions to n. The default is 5.

port=n
Set the server IP port number to n.

noac
Do not cache attributes. This is required when close synchronization with the server is required. Note that using this option will drastically impair performance on the filesystem being mounted.

intr
Allow keyboard interrupts on hard mounts. This allows the user to kill a process that is hung while waiting for a response on a hard-mounted filesystem.

tcp
Use TCP as the transport for this filesystem.

spongy
Set soft semantics for stat, lookup, fsstat, readlink, and readdir (NFS operations and hard semantics for all other NFS operations on the filesystem).

You do not need to specify -f NFS to mount an NFS filesystem. You can specify any filesystem-specific option modifiers using the -o option (only valid options will be applied).


NUCAM
NetWare® Automounter filesystems.

The resource argument is ignored for NUCAM filesystems but mount expects a dummy entry to be present. A suitable placeholder is a repeated directory entry:

/etc/mount -f NUCAM /mount_point /mount_point

There are no NUCAM-specific mount options. The only supported generic options are ro and remount; you cannot mount the filesystem with the rw option.


NUCFS
Volumes on NetWare servers.

Specify the remote resource in the form server:volume where volume must not begin with a slash (/).

All the generic mount options are supported. The following option modifiers are also available:


gid=GID|group
Specify the group name or ID that will be used to open the NetWare volume to be mounted.

grpid
Create files with the GID set to the GID of the parent directory (default).

nogrpid
Create files with the GID set to the effective GID of the calling process.

uid=UID|user
Specify the user name or ID that will be used to open the NetWare volume to be mounted.
The user calling mount (or the requested uid) must be authenticated (via nwlogin) to ``NetWareServer'' for the mount to succeed.

RCKRDG
Rock Ridge CD-ROM filesystems.

The hide, lower and version option modifiers described for the HS filesystem type also apply to Rock Ridge filesystems.

In addition, the following option modifiers are only valid on Rock Ridge filesystems:


dev=cd_path:devnum
Define devnum to be the device number for the device file specified by the pathname on the CD-ROM filesystem. cd_path must be a fully qualified pathname from the root of the CD-ROM. devnum must be a valid device number in the range 0 to 65535.

devmap=pathname
devmap can be used where there would otherwise be several dev mappings on the command line. The file specified by pathname contains lines of the form:

cd_path:devnum [comment]


uid="cd_uid:UID|user"

gid="cd_gid:GID|group"
uid maps the user ID cd_gid recorded on the CD-ROM to a system user ID.

gid is similar to uid but maps a group ID.

If specified, user (group) must be a valid user (group) name; UID (GID) must be a valid user (group) ID in the range 0 to 60000.


uidmap=pathname

gidmap=pathname
uidmap can be used where there would otherwise be several uid mappings on the command line. The file specified by pathname contains lines of the form:

cd_uid:UID|user [comment]

where each line defines a user ID mapping.

gidmap is similar to uidmap but the mapping file has the format:

cd_gid:GID|group [comment]

where each line defines a group ID mapping.


write
Make device files and named pipes writable. (Filesystems mounted with this option modifier are not XPG3 compliant. However, if this option modifier is not given, opening a file on a read-only filesystem will give rise to an EROFS error.)

S51K
AT&T UNIX System V® 1KB filesystems.

The nochkpt and tmp option modifiers described for the AFS filesystem type are also valid with S51K filesystems.


XENIX
XENIX[®] 286/386 filesystems.

umount usage

umount announces to the system that the previous mounted special device or resource is to be removed. Alternatively, the mount point directory can be specified; umount searches the filesystem table /etc/default/filesys to find the corresponding special file. Any pending I/O for the filesystem is completed and the file structure is marked as clean.

Busy filesystems cannot be dismounted with umount. A filesystem is busy if it contains an open file or an active user's working directory.

Filesystem-dependent binaries

The directories /etc/fscmd.d/fstyp contain variations of programs (such as fsck) that depend on the filesystem type; mount and umount invoke the appropriate binary.

Exit values

mount returns the following values:

0
Success.

1
Failed to open mount table.

2
Failed (any other reason).

Diagnostics

mount prints an appropriate diagnostic if the mount fails. A warning is issued if the filesystem to be mounted is already mounted at a different mount point.

Invalid option modifiers produce the following message for a filesystem type fstyp:

mount: invalid fstyp option: invalid_string

A warning is issued if the kernel has not been configured to support the filesystem type to be mounted. Use mkdev(ADM) to configure the kernel to support DOS, HS, and XENIX filesystems.

umount fails if a special device or resource is not mounted or is busy, with the message:

device busy

which indicates that an executing program is using a file on the filesystem to be unmounted. This is often caused by invoking umount while your current working directory is in the filesystem, or by someone else working in the filesystem.

Examples

This example mounts the CD-ROM device /dev/cd0 on /mnt, converts filenames to lowercase, outputs file version numbers, makes device files and named pipes writable, maps device numbers as defined in mapfile, and sets the default user ID to root:

mount -f HS,lower,write,devmap=./mapfile,defuid=root /dev/cd0 /mnt

Using the -o option, this is written as:

mount -f HS -o lower,write,devmap=./mapfile,defuid=root /dev/cd0 /mnt

The following commands are equivalent:

mount -o ro,nosuid,trunc foobar:/remotefs /mnt

mount -f NFS,ro,nosuid,trunc foobar:/remotefs /mnt

mount -f NFS,ro -o nosuid,trunc foobar:/remotefs /mnt

mount -f NFS -o ro,nosuid,trunc foobar:/remotefs /mnt

Warning

Always umount filesystems on floppy disks before removing them from the floppy drive. Failure to do so will require fsck to be run when the disk is next mounted.

Use of the option modifier noac with NFS-mounted filesystems will drastically impair performance on the filesystem being mounted.

Limitations

The mount special device must be a block special file; however, it can be a block special file that has been associated with a regular file by using marry(ADM).

You must have installed NFS to be able to mount a remote filesystem.

Note that when the system is in single-user mode, the commands that look in /etc/mnttab for default arguments (for example bcheckrc(ADM), df(C), ncheck(ADM), quot(ADM), mount, and umount) give either incorrect results (due to a corrupt mnttab(F) left over from a non-shutdown stoppage) or no results (due to an empty /etc/mnttab from a shutdown stoppage).

When in multiuser mode, this is not a problem; the /etc/bcheckrc script initializes /etc/mnttab to contain only /dev/root, and subsequent mounts update it appropriately.

The mount and umount commands use a lock file to guarantee exclusive access to /etc/mnttab. The other commands which read /etc/mnttab do not check the lock file; they may return incorrect values if they are used while a filesystem is being mounted or unmounted. This is not a problem in practice since mount and umount are not frequent operations.

Note that /etc/mnttab is a regular file which can be modified. Because of this, the contents can be inconsistent with the real state of the system. For example, if the file is absent, mount will report that no filesystems are mounted.

Authorization

Only root or a user authorized by asroot(ADM) can use the mount and umount commands to mount or unmount filesystems.

Anyone can use mount to list mounted filesystems and resources.

Files


/etc/mount
mount executable file

/etc/umount
umount executable file

/etc/mnttab
mount table

/etc/default/filesys
filesystem default information

/etc/fscmd.d/fstyp/*
filesystem-dependent programs

See also

default(F), filesys(F), fstyp(ADM), marry(ADM), mkdev(ADM), mnt(C), mnttab(F), mount(S), mountall(ADM), mountd(NADM), nfsd(NADM), setmnt(ADM), showmount(NADM)

Standards conformance

mount is conformant with AT&T SVID Issue 2.
© 2004 The SCO Group, Inc. All rights reserved.