Re: [PATCH 01/19] netfilter: move nf_conntrack initialize out of pernet operations

[ebiederm@xmission.com (Eric W. Biederman)]

canqun zhang <canqunzhang@gmail.com> writes:

> Hi all
> As discussed above,if the host machine create several linux
> containers, there will be several  net namespaces.Resources with "nf
> conntrack" are registered or unregistered on the first net
> namespace(init_net),But init_net is not unregistered lastly,so
> cleanuping other net namespaces  will triger painic.
> If net namespaces are created  with the order of 1,2,...n,they should
> be cleaned with the order of n,...2,1,so in this case init_net will be
> unregistered lastly.

No.  Network namespaces in general can be cleaned up in any order.

In particular you should never ever expect to see the order
n,n-1,n-2,...,2,1.

It may make sense to special case init_net in the cleanup order
but I would really rather not.

Now init_net is special and really should never be cleaned up
for non-modular code.  So it almost makes sense to special
case init_net.

Does anyone know why Alexy decided to do this only for init_net?

My inclination is that Gao Feng is on the rigt path by just removing
the strange init_net special case and performing the work once
per module load, and once per module unload.

> I fixed it up (see below). I have taken a lot of test!

Thank you.

It is nice to see that we have exposed this mis-assumption.

I am inclined to leave the order of this list as is so that
other assumptions of network namespace unregistration order
are exposed.

Unless there is a truly good reason to perform magic on init_net.

Eric

> diff -r 6a1a258923f5 -r 2667e89e6f50 net/core/net_namespace.c
> --- a/net/core/net_namespace.c  Fri Dec 28 11:01:17 2012 +0800
> +++ b/net/core/net_namespace.c  Fri Dec 28 11:05:12 2012 +0800
> @@ -450,7 +450,7 @@
>
>         list_del(&ops->list);
>         for_each_net(net)
> -               list_add_tail(&net->exit_list, &net_exit_list);
> +              list_add(&net->exit_list, &net_exit_list);
>         ops_exit_list(ops, &net_exit_list);
>         ops_free_list(ops, &net_exit_lis
>