Sign up ×
Unix & Linux Stack Exchange is a question and answer site for users of Linux, FreeBSD and other Un*x-like operating systems. It's 100% free, no registration required.

Can i mount a loop file without root permission , normally i do:

mount -o loop DISK_IMAGE FOLDER

Without setting up sudo or suid mount program , is there any suitable way to do this ?

I know i can use fusermount with some iso images , but that is pretty limited , even for ISO images , some of my images can't be mounted , but mount always works

share|improve this question
If your mount binary doesn't require SUID permissions then you should be able to use fakeroot without problems. – ierax Feb 18 '12 at 15:31
@faif fakeroot isn't going to help here: it pretends file ownership is different, but it can't give you permissions that you don't have, such as to call mount(2) when you aren't root. – Gilles Feb 20 '12 at 2:55
@faif Mount absolutely requires root permissions. The potential consequences otherwise would be pretty disastrous. – Shadur May 21 '13 at 20:28

4 Answers 4

up vote 10 down vote accepted

You can't mount anything that the administrator hasn't somehow given you permission to mount. Only root can call the mount system call. The reason for this is that there are many ways to escalate privileges through mounting, such as mounting something over a system location, making files appear to belong to another user and exploiting a program that relies on file ownership, creating setuid files, or exploiting bugs in filesystem drivers.

The mount command is setuid root. But it only lets you mount things that are mentioned in fstab.

The fusermount command is setuid root. It only lets you mount things through a FUSE driver, and restricts your abilities to provide files with arbitrary ownership or permissions that way (under most setups, all files on a FUSE mount belong to you).

Your best bet is to find a FUSE filesystem that's capable of reading your disk image. For ISO 9660 images, try both fuseiso and UMfuse's ISO 9660 support (available under Debian as the fuseiso9660 package).

share|improve this answer

You can use the FUSE module guestmount to mount several types of disk images. It's part part of the guestfs ecosystem and won't require root permissions.

Take a look at the man page for further details.


1. For a typical Windows guest which has its main filesystem on the first partition:

guestmount -a windows.img -m /dev/sda1 --ro /mnt

2. For a typical Linux guest which has a /boot filesystem on the first partition, and the root filesystem on a logical volume:

guestmount -a linux.img -m /dev/VG/LV -m /dev/sda1:/boot --ro /mnt
share|improve this answer

The way possible would be to add an /etc/fstab entry for the ISO with the 'user' parameter, like

/test.iso /mnt/iso auto defaults,user 0 1

But you usually need root access anyway to edit this file, so it's not very helpful.

share|improve this answer

It is actually very easy to mount more or less whatever you want as a normal user without root privileges, provided the right entry has been created in /etc/fstab.

Of course, modifications to /etc/fstab require root privileges. But a single entry can be used with much flexibility to (u)mount many different files on different mount points, without any further editing of /etc/fstab.

Here are two very short (5 lines + comments) Bash scripts that will do the job:

for mounting

# usage: usmount device dir
# author: babou 2013/05/17 on
# Allows normal user to mount device $1 on mount point $2
# Use /etc/fstab entry :
#       /tmp/UFS/drive /tmp/UFS/mountpoint  auto users,noauto 0 0
# and directory /tmp/UFS/
# Both have to be created (as superuser for the /etc/fstab entry)
rm -f /tmp/UFS/drive /tmp/UFS/mountpoint
ln -s `realpath -s $1` /tmp/UFS/drive
ln -s `realpath -s $2` /tmp/UFS/mountpoint
mount /tmp/UFS/drive || mount /tmp/UFS/mountpoint
# The last statement should be a bit more subtle
# Trying both is generally not useful.

and for dismounting

# usage: usumount device dir
# author: babou 2013/05/17 on
# Allows normal user to umount device $1 from mount point $2
# Use /etc/fstab entry :
#       /tmp/UFS/drive /tmp/UFS/mountpoint  auto users,noauto 0 0
# and directory /tmp/UFS/
# Both have to be created (as superuser for the /etc/fstab entry)
rm -f /tmp/UFS/drive /tmp/UFS/mountpoint
ln -s `realpath -s $1` /tmp/UFS/drive
ln -s `realpath -s $2` /tmp/UFS/mountpoint
umount /tmp/UFS/drive || umount /tmp/UFS/mountpoint
# One of the two umounts may fail because it is ambiguous
# Actually both could fail, with careless mounting organization :-)

The directory /tmp/UFS/ is created to isolate the links and avoid clashes. But the symlinks can be anywhere in user space, as long as they stay in the same place (same path). The /etc/fstab entry never changes either.

VITAL WARNING: Mounting is restricted for good security reasons. Making it more flexible may open doors for malicious software. I am not a security expert and I would recommend that you open doors no more than absolutely required ... using options to restrict what can be done with the file systems that can thus be mounted. If a knowledgeable contributor could comment further on security issues, it might be useful.

Various options are available to restrict the use of file systems that are mounted, such as noexec which prevents execution of binaries, or nosuid, and thus contribute to security. Actually, these options are added as default options when the options user or users are used, which is necessarily the case in what we do below. Think twice before you override these defaults.

Other options can be added for further protection. For example, the option owner in the /etc/fstab entry will let users deal only with files or devices they own. See man mount for a list of options:

The use of this /etc/fstab entry can also be restricted through the ownership of the directory (or directories) containing the symlinks.


This explanation was written before I realised I could simplify things to the two scripts above. I did not think of them right away partly because I have at hand a slightly more complex problem that they do not solve without some extra machinery. Thus my explanation may be a bit more intricate than it should, but I do not have the courage to rewrite it all from scratch.

The basic idea is to create entries in /etc/fstab that include the option user or users so that a user can ask mount to do the mounting specified in that entry by giving as argument the file to be mounted or the mount point to use (but not both in my expérience).

You also need a proper entry to umount (which is a slightly different problem - see below). The option user is usually better than users since it restricts permission to umount to the user who mounted the file system, while users will allow that to all. Unfortunately the option user does not always work, and may entail some other steps to be made to work. This is discussed in Option "user" work for mount, not for umount.

First you add to /etc/fstab an entry such as:

/tmp/UFS/drive /tmp/UFS/mountpoint  auto users,noauto, 0 0

and use /tmp/UFS/drive as a symbolic link (or symlink) to whatever device or file you wish to mount, say a file containing the image of an ISO file system /home/johndoe/john-image-file.iso.

You also define /tmp/UFS/mountpoint as a symlink to the mount point you wish to use, say /mnt/iso.

You can then mount john-image-file.iso with the command :

$ mount /tmp/UFS/drive

This is sufficient on my Mageia Linux, since the use of loop devices has now been made implicit, and no longer requires using -o loop explicitly. I do not know how general that is today. See When mounting, when should I use a loop device?

This mounting appears in tables and commands :

$ df | tail -1
/dev/loop0       5,1G  5,1G     0 100% /mnt/iso
$ tail -1 /etc/mtab
/dev/loop0 /mnt/iso udf ro,nosuid,nodev,noexec,relatime,utf8 0 0
$ mount | tail -1
/home/johndoe/john-image-file.iso on /mnt/iso type udf (ro,nosuid,nodev,noexec,relatime,utf8)
$ tail -1 /proc/mounts
/dev/loop0 /mnt/iso udf ro,nosuid,nodev,noexec,relatime,utf8 0 0
$ tail -1 /proc/self/mountinfo
46 22 7:0 / /mnt/iso rw,nosuid,nodev,noexec,relatime - udf /dev/loop0 ro,utf8
$ tail -1 /proc/self/mountstats 
device /dev/loop0 mounted on /mnt/iso with fstype udf

The mounting operation could work for any file or drive and requires only to make a symbolic link from /tmp/UFS/drive to that file or to the device for the drive. Of course, another name and location could be chosen for the symbolic link, as long as it never changes.

Dismounting the file relies in the same way on appropriate use of symbolic links. In the case of a normal device corresponding to some harware drive, you just use the same links.

However, files containing the image of a file system are mounted via a special kind of device called a loop device, automatically allocated when you mount the file.

To dismount the file, you need to refer to the loop device, not the file. Hence you need in /etc/fstab an entry that matches both the loop device used in /etc/mtab, here /dev/loop0, and the mount point, here /mnt/iso.

You cannot create such an entry in advance since the loop device may vary, as they are allocated dynamically. Note that it is also possible to use a fixed loop device, but it is inconvenient in other ways. See (this blog actually inspired the reply here).

However, you can find the name of the loop device, here /dev/loop0, by asking the system, like we did above in several different ways. Then our standard /etc/fstab entry can be made to point to the right loop device via the symlink /tmp/UFS/drive, and to the mount point as done previously with /tmp/UFS/mountpoint. This done, the file may be dismounted with any of the following commands (provided there is no ambiguity with /etc/mtab, which is a different problem):

$ umount /tmp/UFS/drive
$ umount /dev/loop0
$ umount /mnt/iso
$ umount /tmp/UFS/mountpoint

Since the two symlinks are needed only when the commands are issued, they can be changed dynamically. So our single /etc/fstab entry allows mounting any number of files, and umounting them in any order, without root privileges.

Other references:

share|improve this answer
When fstab entries point to symlinks, can't they be exploited to mount anything without root, by just creating symlinks? Indirectly, isn't it just elevating mount command for every normal user?, and so the security issues associated with it? Correct me if I'm wrong. – Bharat G Oct 18 at 4:20
And recent versions of GNU/Linux distros (say debian jessie) will not provide a gui login when it detects an invalid entry in fstab. It falls back to console login session if the device or the mountpoint section of fstab doesn't point to a valid entry. The user has to login via console session and enter startx to explicitly start a display manager. – Bharat G Oct 18 at 4:49

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.