Re: Bug#704744: pbuilder: umounts /{dev,run}/shm of the *host* system
Stéphane Glondu dixit:
>Why not just do nothing if /dev/shm is a symlink?
A $chroot/run/shm should probably be umounted.
>Are there cases where umount_on_exit is called on a symlink that should
>be followed? If not, I would just kill the problem directly there, as in
I was thinking about this a bit, and tonight wondered whether/why
things like readlink/realpath (the shell utilities, not the syscalls
or libc functions) do not offer the functionality of resolving symlinks
(should they encounter any) relative to some (emulated) chroot path.
That would probably help…
Cyril Brulebois dixit:
>Next time, can you please put the right people in the loop?!
>
>Cc-ing:
> debian-bugs-dist@lists.debian.org
I did a reply-to-all on the mail.
>is just plain stupid. Maintainers of the package you're reassigning to
>don't get your control mail. Way to communicate!
I did not know that. I think this is a debbugs bug; it’s inconsistent
to require of the bug submitter to manually look up maintainers (for
example, I wasn’t aware debootstrap has anything to do with booting
Debian…) when reassigning, when one doesn’t have to do so normally.
>Not sure I agree that's an RC bug in debootstrap. I'd rather call it a
If debootstrap tries to umount a symlink that points to outside
the chroot, I’d call it an issue with debootstrap, independent of…
>wishlist in debootstrap to support the new thing pbuilder imposes, and
>an RC bug in pbuilder not to depend on a debootstrap version
>implementing said improved behaviour.
… this one.
Yes, the maintainers of debootstrap (and possibly cdebootstrap),
pbuilder and cowbuilder should probably talk to each other.
Ideally.
But right now, we have a bug that breaks unrelated software,
by umounting my /run/shm every time I create a chroot, and
thus deleting every piece of data that was put there. In this
particular instance, I’m the user who doesn’t really care
about where the bug is, or whose fault it is – this is one
of the major problems with a system based on packages of
separate maintainers…
bye,
//mirabilos
--
Sorry, I’m annoyed today and you came by as an Arch user. These are the
perfect victims for any crime against humanity, like systemd, feminism
or social democracy.
-- Christoph Lohmann on dev@suckless.org
Reply to: