When I run the chroot command an error is given:

failed to run command ‘/bin/bash’: No such file or directory 
  • 1
    Can the question be considered a pure duplicate of unix.stackexchange.com/questions/76490/…? The answers to the questions represent a possible solution for the problem definitely worth a link, but that doesn't make the question a duplicate of it. – Karl Richter Sep 26 '14 at 17:15
  • 1
    The issue for me was that I was using a 32-bit Live CD to mount a 64-bit OS disk and chroot to it. A 32-bit kernel can't run 64-bit bash. The solution was to get a 64-bit Live CD. (The linked duplicate is entirely unrelated.) – Leo Oct 2 '14 at 17:41
  • This is not a duplicate, despite the explanation of the source of the problem being applicable to both questions. The question this is marked a duplicate of is about missing libraries on a generic install, whereas this question is specifically about an error occurring in a chrooted environment. – bschlueter Nov 13 '17 at 22:09

10 Answers 10


This error means that there is no /bin/bash directory inside chroot. Make sure you point it to where bash (or other shell's) executable is in chroot directory.

If you have /mnt/somedir/usr/bin/bash then execute chroot /mnt/somedir /usr/bin/bash.

Apart from the above, you also need to add libc directory dependencies, as mentioned in the answer here.

  • 3
    There is a /bin/bash file in the rootfs folder – USER3254789 May 6 '14 at 7:24
  • 2
    It might be caused by some failing command/line in /root/.bashrc or /root/.bash_profile in your chroot. Can you temporarily rename these files? Also can you make sure that bash is executable (chmod +x /chroot/bin/bash)? – phoops May 6 '14 at 7:28
  • 54
    I figured it out. bin/bash is there, but I didn't have /lib and /lib64 inside it. /bin/bash depends (ofc) on libc, ld-linux, libdl etc... So simple cp -a /usr rootfs/, cp -a /lib rootfs/, cp -a /lib64 rootfs/ was enough. (You can mount-bind those ofc, but I copied them, because I want to run something dangerous, which might corrupt those files in rootfs.) The message from chroot could be more descriptive. "no such file or directory" really means "I can't run this sh...". – Dalibor Filus Nov 3 '15 at 14:28
  • 1
    @EmilVatai added :-) – Dalibor Filus Jan 12 '18 at 11:08
  • 2
    You can run ldd /bin/bash to see the exact files bash depends on that have to be copied. – grabantot Jan 19 '20 at 12:51

I had /bin/bash inside chrooted directory, but I didn't have /lib and /lib64 inside it. The message from chroot could be more descriptive. "no such file or directory" really means "I can't run this...".

/bin/bash depends of course on libc, ld-linux, libdl etc., you can use ldd /bin/bash to see which libraries it requires.

1) You can mount -o bind these directories under chroot 2) Or you can copy these libraries to chroot, if you don't trust the chrooted env to not corrupt them, like so:

cp -a /usr rootfs/
cp -a /lib rootfs/
cp -a /lib64 rootfs/
  • this will create duplicates.. which is not optimized when we have lots of setups – yellowandred Aug 28 '18 at 13:30
  • 1
    This doesn't create duplicates if you use the first method (marked as 1)). The second one is useful if you chroot to untrusted environment. For example, you have a partition with a trojan or something. – Dalibor Filus Aug 30 '18 at 9:43
  • I accidentally made my system unbootable when interrupted cp -a /usr /mnt – Ivan Avdonin Jan 19 '20 at 3:12

chroot tries to start the shell that is set in your $SHELL environment variable by default, but it looks for it in your new root dir, which seems not to contain /bin/bash, so it cannot start.

You can tell chroot to start another program inside the new root by simply adding it as a parameter:

chroot /your/new/root /bin/foo --options...

Note that the path of the command is interpreted inside your new root, so in this example the called program is in fact in /your/new/root/bin/foo

  • 3
    There is a /bin/bash file in the rootfs file so what is the problem – USER3254789 May 6 '14 at 7:18
  • 3
    to whoever downvoted: although this was not the problem in the poster's case, this is a valid and not unprobable explanation of the error in the question. If you see any other problem, please leave a comment when you downvote something. – crater2150 Sep 17 '14 at 19:10

I was getting the same error when trying to ssh to a chrooted account on a remote server. In my case, I was missing the following file in the remote lib64 directory. Server is Centos6.9


It was fixed by executing the following:

cp /lib64/ld-linux-x86-64.so.2 /secure/jail/lib64/
  • didn't fix it for me, but doing cp -r /lib /lib64 /secure/jail fixed it, i needed something from both lib and lib64, and i didn't bother to figure out exactly what. (probably because i had multiarch enabled) – hanshenrik Dec 15 '17 at 21:28

In case you are doing a cross compilation you need to use qemu simulator which can run /mnt/somedir/bin/bash. Below are the steps for armhf cross compilation. Steps for other architectures should be similar.

  1. First install qemu-user-static

sudo apt-get install qemu-user-static

  1. Then copy the 'qemu-arm-static' binary into the chroot directory

sudo cp /usr/bin/qemu-arm-static /mnt/usr/bin/

Once you copy the qemu-arm-static into the /mnt/usr/bin you will be able to do chroot.

Check this out for more details: https://blog.lazy-evaluation.net/posts/linux/debian-armhf-bootstrap.html

  • 1
    There is no indication that this is what the user is trying to do. – Kusalananda Jul 25 '19 at 11:42
  • 2
    The errors are same for both the cases. If someone who is doing cross compilation faces this issue, he can find the answer here. – Jainam MJ Jul 27 '19 at 10:36
  • 1
    That was actually my case, I created a rootfs for arm64 on a x86 and using debootstrap but in order for the chroot to work I had to run qemu-debootstrap which added 'usr/bin/qemu-aarch64-static` under the rootfs. Somehow the chroot command is able to detect it and execute the bash inside the qemu – dafnahaktana Jan 7 '20 at 16:46

you need to run ldd against bash ldd $(which bash), then you might find a missing dependency, for example if you didn't mount/copy lib64, for 64 systems, it will through this error.


Something no one has mentioned yet, if the goal is not to keep copies of libraries you locate with ldd. When you build busybox it respects LDFLAGS=--static per their FAQ. This will build all necessary libraries into your binaries. This does increase the size of the binaries, but... you'd need most of this disk space to store what you're locating with ldd anyway.

Note that you may still need to copy your c library (libc.so.6), core math library (libm.so.6), namespace resolution library (libresolv.so.2), and kernel library (in my case, since I am using a raspberry pi, ld-linux-armhf.so.3) . You can use the ldd tool as directed in other answers on your static busybox binary to discover whether this is the case.

These may in turn depend on other libraries. To discover whether this is the case you can use the file tool. I am using the full path to raspberry pi's libm.so.6 as an example :

file /chroot/lib/arm-linux-gnueabihf/libm.so.6

In my case, since ARM processors need many libraries, I copied my entire arm-linux-gnueabihf folder into my lib folder, allowing me to access my chroot.


The bash elf is not available in /bin, /usr/bin/ etc.

cp -rf --preserve=links  /usr/bin/* myroot/usr/bin/; cp --preserve=links  -rf /lib64/*  myroot/lib64/;..

copy all libraries and ELFs.

  1. Run ldd /bin/bash to see dependences. Wiil reply like follows:
linux-vdso.so.1 (0x00007ffd057eb000)
libtinfo.so.6 => /lib/x86_64-linux-gnu/libtinfo.so.6 (0x00007f83183bc000)
libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007f83183b6000)
libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f83181c4000)
/lib64/ld-linux-x86-64.so.2 (0x00007f8318536000)
  1. Then for each run diff. Like diff /lib/x86_64-linux-gnu/libtinfo.so.6 /mnt/lib/x86_64-linux-gnu/libtinfo.so.6 etc. (Change /mnt to your mount path.)

    Also run diff for the bash, like diff /usr/bin/bash /mnt/usr/bin/bash.

    Skip linux-vdso.so.1 because it inserted into kernel.

  2. If will found differences then make replace these by files from Live CD.

  3. Also check that the symlink links /lib, /lib32, /lib64 lead to the correct directories (to /usr/lib etc).

sudo nano /etc/passwd 

After that, look for the user that gives you the problems and verify that in front of / bin / bash there are no spaces or any other character, in case of chroor, verify the root user.

nano can be replaced by any text editor.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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