Comment 10 for bug 1401148

So for now I added also a task for the kernel, though the truth (if such a thing exists) could be somewhere between. Serge, Stephane, what we probably need to figure out is what exactly lxc-start tries to get done when slave mounting /run/netns. And somehow it might be possible that it needs improvement for the case that this is denied or fails. Looking at it from the outside it feels like going on assuming it got its own space but actually continuing to use the host space.
The other thing would be that this sound like lxc-start would require a rule to actually allow it to do that mount of /run/netns.