mount, umount - Mount or unmount a file system
caddr_t data ); int umount(
int umnt_flag );
Defines the type of the file system. The types of recognized file sys‐
tems are: Reserved for third-party file systems. See NOTES for informa‐
tion about support for third-party file systems. For internal use
only. For internal use only. For internal use only. Compact Disk
File System (see cdfs(4)) Distributed File System (layered product)
Versatile Disk File System (see dvdfs(4)) DCE Episode File System (lay‐
ered product) File Descriptor File System (used by streams) File on
File Mounting File System (used by streams) Memory File System (RAM
disk) Advanced File System (AdvFS) Network File System, Version 2 pro‐
tocol Network File System, Version 3 protocol PC File System /proc File
System (used by debuggers) System V File System Berkeley's UNIX File
System Points to a null-terminated string that contains the appropriate
pathname. Specifies which semantics should be used when accessing the
file system. One or more of the following flags might be valid, depend‐
ing on the file system type and flag combination: Cause all files in
the mounted AdvFS fileset to use atomic-write data logging. (See the
description of the adl argument for the mount command's -o option in
mount(8).) For internal use only. For UFS, flush data asynchronously
rather than synchronously. For information about the advantages and
risks of using this flag, see the discussion of the delayed keyword for
the mount command's -o option in mount(8). For internal use only.
Allow an AdvFS fileset to be mounted as a domain volume even though it
has the same AdvFS domain ID as a fileset that is already mounted.
Allow the file system to be exported for both read and write access.
Allow the file system to be exported for read-only access. For UFS and
AdvFS, extend the size of the file system to use all the available
storage space in a revised partition. The file system must be already
mounted in order to use this option. For internal use only. Forcibly
mount the file system, even if it is unclean. In a cluster, enable
cluster partitioning, which restricts use of the file system to the
member that mounts it. This flag cannot be used on a file system that
is already mounted.
This flag is automatically set when mounting a UNIX file system
(UFS) for read-write access and when mounting an in-memory file
system (MFS). For internal use only. All new files and direc‐
tories inherit the group ID of the parent directory.
When this flag is not specified, the following SVID III seman‐
tics apply: If the parent directory's mode bits include the
IS_GID, then the group ID of the new file or directory is the
parent directory's group ID. If the parent directory's mode
bits do not include IS_GID, then the group ID of the new file or
directory is the process group ID of the creating process. For
internal use only. For internal use only. Obsolete; not used.
For internal use only. Mark the file access time changes made
for reads of regular files in memory, but do not flush them to
disk until other file modifications occur. This behavior does
not comply with industry standards and is used to reduce disk
writes for applications with no dependencies on file access
times. Do not allow access from the file system to either
block- or character-special devices. Do not allow files to be
executed from the file system. Do not honor setuid or setgid
bits on files when executing them. For AdvFS and UFS, enable
quotas on the file system. The file system should be treated as
read only; no writing is allowed (even by a process with appro‐
priate privilege). Physically write-protected and magnetic tape
file systems must be mounted read only or errors will occur when
access times are updated, whether or not any explicit write is
attempted. Obsolete; not used. For AdvFS and UFS, enable an
alternate smooth sync policy wherein dirty UBC pages are flushed
to disk after the smoothsync_age period, but only if they are
idle for the smoothsync_age period. By default, dirty UBC pages
are written to disk after the smoothsync_age period, regardless
of whether they are still being modified.
This policy can be applied only to dirty pages in the file sys‐
tem cache (UBC); dirty pages mapped into virtual memory are
always flushed to disk after the smoothsync_age period, even if
they are still being modified.
The smoothsync_age system attribute can be configured by means
of the /sbin/sysconfig command. See sys_attrs_vfs(5) and syscon‐
fig(8) for information about the smoothsync_age attribute and
/sbin/sysconfig command, respectively. For AdvFS and UFS, cause
all writes to be written to disk as well as to the buffer cache
before the function performing the write operation returns. By
default, write operations to disk are done asynchronously of
write operations to the buffer cache. For AdvFS and UFS, pre‐
vent excessive asynchronous I/O from overloading the device
queue. This flag has no effect if M_SYNCHRONOUS is applied to
the file system. For internal use only. See M_THROTTLE. The
mount operation is being performed on an already mounted file
system. This flag allows mount attributes to be changed without
unmounting and remounting the file system.
The attributes that can be changed for a mounted file system are
restricted by most types of file system software. For example,
for most types of file systems, you cannot change the access
mode from read-write to read-only if the file system is already
For UFS or AdvFS, M_UPDATE is typically specified without
M_RDONLY to change a file system that had been mounted read-only
to read-write. If M_UPDATE is used in a cluster environment, it
is important to remember that while AdvFS filesets can be
mounted read-write and be accessible to all cluster members, UFS
file systems must be mounted read-only to be available to all
cluster members. For UFS, any attempt to use M_UPDATE on a file
system that is already mounted read-only and accessible to all
cluster members will fail. Points to a structure that contains
the type-specific parameters to mount. May be 0 (zero) or the
following: Performs a fast unmount that causes remote file sys‐
tems to be unmounted without notifying the server.
Except in the case of file-on-file mounting, the mount() function
mounts a file system on the directory pointed to by the mnt-path param‐
eter. Following the mount, references to mnt-path refer to the root of
the newly mounted file system.
The mnt-path parameter must point to a directory or file that already
For file-on-file mounting, the mount() function mounts a file specified
by the data parameter onto another file specified by the mnt-path
parameter. The file specified by the data parameter cannot be a direc‐
tory file; otherwise either file may be of any type. The mnt-path can‐
not already have a file system or another file mounted on it.
The umount() function unmounts a file system mounted at the directory
pointed to by the mnt-path parameter. The associated directory reverts
to its ordinary interpretation.
Except for file-on-file mounting and Network File System (NFS) mounts,
to call either the mount() or umount() function, the calling process
must have superuser privilege.
Two mount() functions are supported by Tru64 UNIX: the BSD mount() and
the System V mount(). The default mount() function is the BSD mount()
documented on this reference page.
The operating system does not support the System V lmount() function.
Third-party file systems do not have type constants defined in the
<sys/mount.h> file. For these file systems, functionality has been
added to the mount() function to allow an application to query by using
the file system's name string to obtain the corresponding type numeric
value. The type numeric value obtained from the first mount() call can
then be used in a second mount() call to mount the third-party file
To use the type query functionality, call mount() with type as -1, mnt-
path as NULL, mnt_flag as 0, and data pointing to the address of a
vfsops_fsname_args structure. This structure is defined in the
<sys/mount.h> file and contains two fields; the first field must be set
to the file system name string to search for and the second field is a
return index. If the specified name string is found, the function
returns the corresponding type numeric value into the structure's
return index field.
The mount() function supports mount-point argument pathnames of up to
MNAMELEN, which includes the null terminating character. MNAMELEN can
be up to 90 characters long, including the null terminating character.
The mount() function returns 0 (zero) if the file system was success‐
fully mounted. Otherwise, -1 is returned. The mount can fail if the
mnt-path parameter does not exist or is of the wrong type.
For AdvFS, the mount can fail if the domain or fileset (or both) speci‐
fied in the data parameter does not exist or is inaccessible.
For UFS, the mount can fail if the special device specified in the
ufs_args structure is inaccessible, is not an appropriate file, or is
already mounted. The same failure can occur for a PROCFS file system.
A mount can also fail if there are already too many file systems
mounted, either system wide or for a specific file system type.
For the query by name functionality (for third-party file systems), the
mount() function returns 0 (zero) if the file system name was found;
otherwise, -1 is returned.
The umount() function returns 0 (zero) if the file system was success‐
fully unmounted. Otherwise, -1 is returned. The unmount will fail if
there are active files in the mounted file system, unless the MNT_FORCE
flag is set and the file system supports forcible unmounting.
If the mount() function fails, errno may be set to one of the following
values: The caller does not have appropriate privilege. A component of
a pathname exceeded NAME_MAX characters, or an entire pathname exceeded
PATH_MAX characters. Too many symbolic links were encountered in
translating a pathname. No space remains in the mount table. The file
system is invalid or not installed. A component of the mnt-path param‐
eter does not exist. The specified mnt-path is not a directory. A
pathname contains a character with the high-order bit set, or the file
system name in the query by name functionality is invalid. Another
process currently holds a reference to the mnt-path parameter. The
file system is not clean and M_FORCE is not set. The mnt-path parame‐
ter points outside the process' allocated address space. The process
is attempting to mount on a multilevel child directory.
The following errors can occur for a UFS file system mount: The fspec
field is not a block device. The major device number of fspec is out
of range (this indicates no device driver exists for the associated
hardware). The device pointed to by the fspec field is already
mounted. The superblock for the file system had a bad magic number or
an out-of-range block size. Not enough memory was available to read
the cylinder group information for the file system. An I/O error
occurred while reading the superblock or cylinder group information.
The fspec field points outside the process' allocated address space.
The following errors can occur for an NFS-compatible file system mount:
NFS timed out trying to contact the server. Some part of the informa‐
tion described by nfs_args points outside the process' allocated
The following errors can occur for a PROCFS file system mount: The
device pointed to by the fspec field is already mounted. No space
remains in the mount table. The fspec field points outside the
process's allocated address space. Free vnodes are not available.
Memory is insufficient for the /proc directory table.
If the umount() function fails, errno may be set to one of the follow‐
ing values: The caller does not have appropriate privilege. A compo‐
nent of the path is not a directory. The pathname contains a character
with the high-order bit set. A component of a pathname exceeded
NAME_MAX characters, or an entire pathname exceeded PATH_MAX charac‐
ters. Too many symbolic links were encountered in translating the
pathname. The requested directory is not in the mount table. A
process is holding a reference to a file located on the file system.
An I/O error occurred while writing cached file system information.
The mnt-path parameter points outside the process' allocated address