Multiple mounts of chroot/dev, chroot/run cause 'cannot execute 'command': unable to open pty'

[rebast85]

** This appears to be a duplicate of #48. Either way I fixed it in a decent manner.

Describe the bug
Multiple mounts of dev and run in the chroot file system cause the host system to give the error ''cannot execute 'command': unable to open pty'.
For example, when trying to run 'sudo umount chroot/run', one gets the error.
This happens in both the GUI terminals as in the linux command line itself (the one reachable with ctrl+alt+F3 etc)

** NOTE:
I think this may only be related to the /dev mount in the chroot but I am not sure.
The reason I thought /run had to do anything with it was because /dev and /run are the only ones in the chroot getting mounted multiple times when rerunning the script.
Having multiple mounts of such directories onto the same path is a bad idea whatsoever.

This happens because the script does not unmount the mounted directories in the chroot when it exits because of an error. Currently the error in the chroot_build.sh script is due to wireless-tools not being available. I will file another bug report for this issue.

The only way to get the host system to work again is to perform a reboot.

To Reproduce
Steps to reproduce the behavior:

1. Run the build.sh script. Once it calls chroot_build and an error occurs in there, it exits.
2. Re-run the build.sh script with the chroot_build command. Do this two, three times.
3. Try to run sudo umount chroot/dev or a sudo su.
4. See error.

Expected behavior
I expected this script not to blindly mount already mounted directories. Perhaps a check to see if these are already mounted would be a good idea. Or better, a routine where the script cleans up the mounts when it quits due to an error.

Desktop (please complete the following information):

• OS: debian trixie 13, running the script on an ubuntu 26.04 install in a qemu vm.

Additional context
I will attempt to work in both checks for the mounted filesystems and a routine that cleans up the mounts after an error into your script in the upcoming days when I have some time to do so. I am willing to share the improved bits of code with you.

[rebast85]

I have added code to trap errors and unmount chrooted dirs when they are actually mounted.

I have added the following, between DATE=... and function help()...

<--- DATE=`TZ="UTC" date +"%y%m%d-%H%M%S"`

function error_cleanup() {
    local exit_code=$?
    printf 'Cleanup triggered (exit code: %s)\n' "$exit_code"
    chroot_exit_teardown
}

function help() { --->

I have added a checker function to see if mounts are actually mounted, and adapted chroot_teardown_exit:
This rewrite has the added benefit that if a mount point does not exist, the script will continue to attempt to unmount the others, instead of erroring and stopping and leaving some mounts still mounted.

function safe_umount() {
    local path="$1"

    if mountpoint -q "$path"; then
        sudo umount -l "$path"
    fi
}

function chroot_exit_teardown() {
    safe_umount "${SCRIPT_DIR}/chroot/proc"
    safe_umount "${SCRIPT_DIR}/chroot/sys"
    safe_umount "${SCRIPT_DIR}/chroot/dev/pts"
    safe_umount "${SCRIPT_DIR}/chroot/dev"
    safe_umount "${SCRIPT_DIR}/chroot/run"
}

I have added the following code right above the # ============= main ================ part:

# =============   error trap  ==========

trap error_cleanup EXIT

# =============   main  ================

With these changes in place, the script now unmounts the mounts in the chroot directory when something messes up, for example when the run_chroot part messes up. This makes sure that these system directories do not get mounted multiple times over eachother, causing errors like the 'unable to open pty' one I encountered.

[rebast85]

Notice: I have forked your repo and have started to learn how to use git. I am in the process of improving the script and fixing a few bugs and have added some extra features to it like an error catcher and a function to clean up chroot mounts after a crash. I will send a pull request once I am done with this phase - if that is okay with you. Please check out my repository for a file called 'changes.txt', which lists all the changes.