[NETNS] Oops in register_pernet_operations() with CONFIG_NET_NS=n

[Eric W. Biederman]

"Denis V. Lunev" <dlunev at gmail.com> writes:

> Eric W. Biederman wrote:
>> Benjamin Thery <benjamin.thery at bull.net> writes:
>> 
>>> Eric W. Biederman wrote:
>>>> Benjamin Thery <benjamin.thery at bull.net> writes:
>>>>
>>>>> Denis V. Lunev wrote:
>>>>>> The patch attached should help. The idea is simple. The "init" should be
>>>>>> called only once without NETNS. Period. No need for any lists.
>>>>> This is the kind of idea I had but I didn't think it could be 
>>>>> that simple. :) 
>>>>> Thanks Denis.
>>>> It isn't.
>
> this will work due to INIT_LIST_HEAD with circles list to itself and a
> del operation will work.

Suppose I have this fragment of code in a module:

> static int __net_init xt_net_init(struct net *net)
> {
>         ...
> }
> 
> static void __net_exit xt_net_exit(struct net *net)
> {
>         ...
> }
> 
> static struct pernet_operations __net_initdata  xt_net_ops = {
> 	.init = xt_net_init,
> 	.exit = xt_net_exit,
> };
> 
> static int __init xt_init(void)
> {
> 	return register_pernet_subsys(&xt_net_ops);
> }
> 
> static void __exit xt_fini(void)
> {
> 	unregister_pernet_subsys(&xt_net_ops);
> }
> 
> module_init(xt_init);
> module_exit(xt_fini);

What happens during module removal when unregister_pernet_subys calls 
xt_net_ops.exit after xt_net_ops has been removed from the kernels
memory?


> By the way, I think that we can in the case of undefined CONFIG_NET_NS
> reduce register to calling ->init method and unregister to calling
> ->exit method.
>
> This is a correct thing at least for now and will be welcomed by the all
> embedded/etc people.

I'm not fundamentally opposed.  Earlier versions of my patchset
did that and more.   However I think the pain is greater then the
gain right now.  Especially since this concept seem to require
having quality inspected into it.

Eric