[Devel] Re: [RFC][PATCH] Do not set /proc inode->pid for non-pid-related inodes

[Serge E. Hallyn]

Quoting Eric W. Biederman (ebiederm at xmission.com):
> "Serge E. Hallyn" <serue at us.ibm.com> writes:
> 
> >> >
> >> > Except that unless we mandate that pid1 in any namespace can't exit, and
> >> > put that feature off until later, we can't not address it.
> >> 
> >> What if we mandate that pid1 is the last process to exit?
> >
> > I think people have complained about that in the past for application
> > containers, but I really don't see where it hurts anything.
> >
> > Cedric, Herbert, did one of you think it would be bad?
> 
> Sure.  As an extension I don't have a problem with the notion, of
> allowing pid1 to exit before others.  But if it makes things harder
> on us I don't want to support it, at least not initially.

So how do you see us enforcing pid1's existance?  Somehow keep it from
fully exiting, or just kill all the processes in it's namespace if it
exits?

-serge
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers