Normal filesystem images can be easily mounted:

mount system.img /mnt

Examined, and modified. But when I try this with a Raspberry Pi system image (e.g. raspbian), I get:

mount: unknown filesystem type '(null)'

And no matter what I try with -t, it won't work. How can I mount this image?

up vote 54 down vote accepted

This question is answered as part of the answer to other questions, but it deserves canonical treatment here so it does not have to keep being repeated.

You can't mount the image as a whole because it actually contains two partitions and a boot sector. However, you can mount the individual partitions in the image if you know their offset inside the file. To find them, examine the image as a block device with fdisk -l whatever.img. The output should include a table like this:

Device         Boot     Start       End  Blocks  Id System
whatever.img1            8192    122879   57344   c W95 FAT32 (LBA)
whatever.img2          122880   5785599 2831360  83 Linux

These are the two partitions. The first one is labelled "FAT32", and the other one "Linux". Above this table, there's some other information about the device as a whole, including:

Units: sectors of 1 * 512 = 512 bytes

We can find the offset in bytes by multiplying this unit size by the Start block of the partition:

  • 1st partition 512 * 8192 = 4194304
  • 2nd partition 512 * 122880 = 62914560

These can be used with the offset option of the mount command. We also have a clue about the type of each partition from fdisk. So, presuming we have directories /mnt/img/one and /mnt/img/two available as mount points:

mount -v -o offset=4194304 -t vfat whatever.img /mnt/img/one
mount -v -o offset=62914560 -t ext4 whatever.img /mnt/img/two

If you get an "overlapping loop" error here, your version of mount requires you to specify the size as well as the offset of the first partition. Unmount that, and use the number of blocks (57344) * 512 (= 29360128):

mount -v -o offset=4194304,sizelimit=29360128 \
    -t vfat whatever.img /mnt/img/one  

The second partition doesn't need a sizelimit since there's nothing after it in the image.

You can now access the two partitions. If you do not intend to change anything in them, use the -r (read-only) switch too. If you do change anything, those changes will be included in the .img file.

Note that the first partition is probably mounted on /boot in the second partition when the system is running.

  • 9
    kpartx makes it even easier for you: kpartx -a whatever.img makes the appropriate (loop) nodes, which you can later mount at will. Run kpartx -d whatever.img to delete the (unmounted) loop nodes. – user29510 Dec 28 '15 at 11:39
  • How to mount root and boot partition simultaneously : unix.stackexchange.com/q/342463/9689 ? – Grzegorz Wierzowiecki Feb 4 '17 at 10:35
  • 1
    P.S. If one want's to mount both partitions at same time, in order to avoid "overlap" error, just specify size of partition with sizelimit option. For more info: unix.stackexchange.com/a/342466/9689 . (Btw. above answer may be updated to provide example usage of that option) – Grzegorz Wierzowiecki Feb 4 '17 at 21:56
  • 2
    Rather than multiplying manually, you can use bash's Arithmetic Expansion: $((8192*512)). – Ruslan Aug 8 '17 at 13:46

losetup provides partition probing through -P. Using this makes mounting partitions of a full disk image such as the Raspbian SD card image very easy:

losetup -P /dev/loop0 raspbian.img
mount /dev/loop0p2 /mnt
mount /dev/loop0p1 /mnt/boot
  • This seems less prone to human error. Can I also use labels in the mount commands? – ctrl-alt-delor Feb 1 at 11:26
  • Very nice, but mostly overseen command! – not2qubit May 11 at 20:09

In case fdisk -l shows up with sectors instead of blocks on Arch linux... Another way to find the number of blocks in case of the "overlapping loop error" is to subtract the offset in bytes of the first partition from the second partition and divide by two. For example (62914560-4194304)/2 = 29360128

Your Answer

 
discard

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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