Cannot delete device /dev/loop0

Question

I unsafely removed a USB device that was attached to loop0 with losetup and could not delete nor detach loop0 afterwards.

losetup -a shows /dev/loop0: [0005]:145606719 (/dev/sdb1)

When I remount the device to /dev/sdb1, umount it and try losetup -d /dev/loop0, I will still get the same error loop: can't delete device /dev/loop0: Device or resource busy. Is there a way to get rid of the loop device?

That's odd, you don't usually attach loop to usb devices. How did you do that? — Keith, May 04 '11 at 09:15
I was using LUKS encryption and thought I could encrypt my 1gb usb flash drive. — BiX, May 05 '11 at 01:30
You can use LUKS for that, but not by using loop device. A LUKS partition has its own partition type. — Keith, May 05 '11 at 02:37
Sorry noobness on my part, cannot find any clear instructions online. — BiX, May 05 '11 at 03:20

score 58 · Answer 1 · edited May 21 '15 at 21:30

58

You may have to also use dmsetup to remove the device mapping. Easiest way, if it doesn't interfere with any other mappings is to use a dmsetup remove_all.

edited May 21 '15 at 21:30

A.D.

4,487
3
38
50

answered Jun 21 '12 at 00:38

Aaron Flin

589
1
4
2

7

This worked for me, though I used `dmsetup info` to get the mappings, then used `dmsetup remove `. Thank you. – unpythonic Aug 07 '13 at 20:17
7

`sudo dmsetup info` gives me `No devices found` – Jayen Feb 12 '14 at 23:21
2

This also worked for me. For some reason using kpartx -a and then kpartx -d on CentOS 6.5 doesn't remove the mappings like it should. – Peter Jul 18 '14 at 06:35
1

it saves the day if you remove loop dev just by simple `rm /dev/loop` without proper `losetup -d` remove before that. – Lapsio May 29 '18 at 22:55

score 18 · Accepted Answer · edited Jan 21 '18 at 03:31

18

Are you sure that the device isn't busy? Have you ever tried fuser to determine an possible PID?

Precautionary get all possible information:

fuser -c /dev/loop0
fuser -d /dev/loop0
fuser -f /dev/loop0

Try to stop the process, which could use /dev/loop0. If necessary use kill -9 or try fuser -k to send the kill signal -> look at man fuser.

edited Jan 21 '18 at 03:31

Peter Mortensen

30,738
21
105
131

answered May 04 '11 at 10:52

The Bndr

13,204
16
68
107

3

What is `fuser -d` or `fuser -f` ? My fuser does not have those flags (`fuser (PSmisc) 22.20` - from Centos 7.6) – Popup Jun 23 '20 at 11:22

score 12 · Answer 3 · edited Aug 07 '14 at 00:04

12

I had a similar issue with an SD card and Aaoron Flin's suggestion to use dmsetup worked for me.

Specifically, you should be able to ls /dev/mapper to see if any loop0pX partitions exist.

If so, you can use dmsetup remove /dev/mapper/loop0p2 to get rid of any unnecessary partitions. This can be useful if you want to keep some mappings attached.

edited Aug 07 '14 at 00:04

Catskul

17,916
15
84
113

answered Jan 11 '13 at 03:44

Mike.P

121
1
2

score 9 · Answer 4 · edited Jan 21 '18 at 03:34

I had the same issue today and none of the previous answers fixed it (I didn't try to reload the loop kernel module, because I wanted to understand the actual issue).

It turns out the image file associated with the loop device contained an "Linux LVM" partition, which was automatically set as shown by pvscan:

$ pvscan
  PV /dev/sda1    VG server-vg   lvm2 [417,76 GiB / 0  free]
  PV /dev/loop0   VG vbox-vg     lvm2 [7,81 GiB / 0    free]
  Total: 2 [425,57 GiB] / in use: 2 [425,57 GiB] / in no VG: 0 [0   ]

So I had de deactivate all the logical volumes inside the volume group:

$ vgchange --activate n vbox-vg
  0 logical volume(s) in volume group "vbox-vg" now active

Eventually I was able to nicely detach the loop device:

$ losetup -d /dev/loop0
$ losetup -a

score 7 · Answer 5 · edited Dec 22 '18 at 19:22

7

I chased this loop solution to the end of the Internet and found the solution is uninstall snapd and purge all associated files: (In my case this was 167 Gb)

sudo apt purge snapd

edited Dec 22 '18 at 19:22

Jignesh Joisar

13,720
5
57
57

answered Dec 22 '18 at 19:06

dwb.kelmark

71
1
1

I purged snapd and the loop device is still. Device: /dev/loop1 (read-only) Contents: Unknown (squashfs 4.0). It happened after I installed Clementine software. I uninstalled the software and this loop device came up. It also says Backing File: /var/lib/snapd/snaps/clementine_274.snap (deleted) – Gautam Sreekumar Mar 12 '19 at 08:10
2

This causes all software installed through snapd to be removed. – wajdi_jurry Jan 08 '21 at 16:34

score 5 · Answer 6 · answered Dec 20 '19 at 04:14

5

Run

mount | grep /dev/loopN

to check if the loop device is being mounted somewhere, then run

umount /dev/loopN

to unmount it.

If still have problem, then run

lsof | grep loopN

to check which process is using it then kill it.

answered Dec 20 '19 at 04:14

osexp2000

2,910
30
29

score 3 · Answer 7 · edited Jan 21 '18 at 03:32

3

Try reloading the loop kernel module. If it does not help then reboot.

edited Jan 21 '18 at 03:32

Peter Mortensen

30,738
21
105
131

answered May 04 '11 at 09:15

tamasd

5,747
3
25
31

2

This actually worked for me (other solutions didn't) – Tuinslak Jul 04 '16 at 12:11
3

For those who aren't sure how to do this: `modprobe -r loop && modprobe loop`. :) – XtraSimplicity Oct 14 '19 at 22:48
modprobe loop is working perfectly if no loop device file is mounted or configured with crypto – Znik Jul 21 '21 at 08:17

score 2 · Answer 8 · answered Jun 10 '16 at 23:42

2

If you're using Luks, you need to luksClose first

cryptsetup luksClose $whatever
losetup -D

answered Jun 10 '16 at 23:42

choad rocker

181
5

score 2 · Answer 9 · edited Feb 22 '19 at 14:40

THX for this article BiX and all: I'm not a Linux administrator only a user but can contribute so much to a had loop device problem:

When a virtual mounted image file (.iso) is not really cleanly removed from the file explorer (Dolphin/Nautilus).

You can unmount this virtual partition using the Prg "Disks" (under Service Prg's the disk management tool). However, the ISO file will still remain as a loop in the device management list. For example, if you want to delete this ISO file reel from the disk, the deletion process will only take place virtually, and the corresponding storage space in MB or GB will not be free at all. The ISO file is no longer visible in Dolphin/Nautilus after the deletion, but the corresponding storage space requirement is not specified as free again; i.e. the information of the free and occupied storage space on the data carrier where this ISO file was is unchanged before and after the deletion.

A simple but unconventional solution is to reboot/start the computer. But that can‘t not be the real solution!

The Linux shell losetup, which is described in the man page, provides a remedy here:

losetup is used to associate loop devices with regular files or block devices, to detach loop devices and to query the status of a loop device. If only the loopdev argument is given, the status of the corresponding loop device is shown.

So you can use losetup -d /dev/loop0 to delete a file associated with a loop on dev/loop0(! "losetup -d" is just usable as admin., so you need to type sudo losetup -d or wath the most LinuxGurus not fink fin, u have a so-terminal static open).

In the man page of losetup you can see that the expression Delete is not used here, but Solve. Nevertheless, according to my example, that a "file is not deleted reel, despite clean unmount" after using the option -d, the file is only now deleted reel. (I'm not a Linux administrator but:) This is rather associated with an assignment of the loopNumber which is virtual, but becomes a real condition when mounting the iso-image (image).

Here the complete syntax to the shell losetup:

https://linux.die.net/man/8/losetup

(I work on a bit corrupted linuxMint-18)

Translated from German with www.DeepL.com/Translator

score 0 · Answer 10 · answered Sep 19 '16 at 20:29

0

Definitely there is a process using it. If you couldn't determine which process is preventing it from deleting, then restart your server if you can.

answered Sep 19 '16 at 20:29

user3578181

1,896
2
17
12

score 0 · Answer 11 · edited Jan 21 '18 at 03:35

0

Same as by @frntn.

Loops were occupied by RAID:

# cat /proc/mdstat

Personalities : [raid0]

md126 : active raid0 loop6p5[1] loop5p3[0]
  209584128 blocks super 1.2 512k chunks

md127 : active raid0 loop6p2[1] loop5p2[0]
  31440896 blocks super 1.2 512k chunks

I removed them and everything obeys the orders again.

edited Jan 21 '18 at 03:35

Peter Mortensen

30,738
21
105
131

answered Feb 11 '17 at 02:51

MolbOrg

105
1
5

score -5 · Answer 12 · answered Sep 15 '17 at 11:23

-5

I just right click the loop device in dolphin and unmount it.

answered Sep 15 '17 at 11:23

Brons0nmarlotcheechdexter

11

Cannot delete device /dev/loop0

12 Answers12