Bug#818978: systemd crashes in lxc on container stop

[0x2a]

to reproduce on a Debian unstable system:

$ sudo apt-get install lxc
$ sudo usermod --add-subuids 100000-165536 $USER
$ sudo usermod --add-subgids 100000-165536 $USER
$ sudo chmod +x $HOME
$ mkdir .config/lxc

create ~/.config/lxc/default.conf with:

lxc.id_map = u 0 100000 65536
lxc.id_map = g 0 100000 65536

$ echo "kernel.unprivileged_userns_clone=1" | sudo tee 
/etc/sysctl.d/80-lxc-userns.conf
$ sudo sysctl --system

$ lxc-create -t download -n systemd-test -- -d debian -r jessie -a amd64
$ lxc-start -n systemd-test
$ lxc-ls -f
NAME         STATE   AUTOSTART GROUPS IPV4 IPV6
systemd-test RUNNING 0         -      -    -

follow the systemd logs inside the container:
$ lxc-attach -n systemd-test
root at systemd-test:/# journalctl -f

while executing lxc-stop on the host:
$ lxc-stop --nokill -n systemd-test


you will see this in the log:

Jul 14 17:11:40 systemd-test systemd[1]: Starting Halt...
Jul 14 17:11:40 systemd-test systemd[1]: Assertion 
'sd_id128_randomize(&id) >= 0' failed at ../src/core/dbus.c:688, 
function bus_on_connection(). Aborting.
Jul 14 17:11:40 systemd-test systemd[1]: Caught <ABRT>, dumped core as 
pid 135.


also the container will still be running if lxc-stop is executed with 
--nokill to request clean shutdown (else it kills the processes after a 
timeout):

$ lxc-ls -f
NAME         STATE   AUTOSTART GROUPS IPV4 IPV6
systemd-test RUNNING 0         -      -    -