[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#982422: debootstrap: segmentation fault around ldconfig during debootstrapping bullseye/arm64 on a buster/amd64 host



> I'm not a debootstrap maintainer, but if you're using QEMU user mode
> to run 
> foreign chroots, I suggest you use the qemu-debootstrap wrapper which
> comes 
> with the qemu-user-static package.

Yes, I think that's what it does implicitly due to some magic
mechanism. I can also explicitly write 'qemu-aarch64-static' in front
of '/bin/bash' in my script, but that seems to do the very same thing.

> If you're using systemd, there's also a handy tool used for entering
> chroots 
> called systemd-nspawn you may be interested in. It takes care of the
> mount 
> points and the other little details. You could just use it like
>       sudo systemd-nspawn -D ./bullseye

Interesting indeed. But do you think it could be related? Well, I'll
give it a try tomorrow. Interesting improvement anyway.

> There's also a slimmer arch-chroot tool in the arch-install-scripts
> package 
> (despite the name, it's useful aside from just Arch Linux chroots).

I would not like to switch to just another tool at that moment. My
actual script is a bit larger and would not be trivial to migrate.

Something seems to break regarding libc-bin, so is this about glibc in
some way? Sorry, I'm not deeply experienced in those system levels. :)
What would be good ways to get more information? I tried with strace
meanwhile, but this is just telling me "PTRACE_TRACEME: Function not
implemented". :-/

As I said, I definitely had it working here, one or two month ago,
without changes on the script. So I'm optimistic that it could be
solvable.

>  Your script ran okay and didn't segfault from my Bullseye host.

That's great to hear. So even if I cannot get it solved, it should go
away in some months. :) Not perfect, but not worst case...


Reply to: