[PATCH] netns: Coexist with the sysfs limitations

[Eric W. Biederman]

Benjamin Thery <benjamin.thery at bull.net> writes:

> Serge E. Hallyn wrote:
>> Quoting Eric W. Biederman (ebiederm at xmission.com):
>>> To make testing of the network namespace simpler allow
>>> the network namespace code and the sysfs code to be
>>> compiled and run at the same time.  To do this only
>>> virtual devices are allowed in the additional network
>>> namespaces and those virtual devices are not placed
>>> in the kobject tree.
>>>
>>> Since virtual devices don't actually do anything interesting
>>> hardware wise that needs device management there should
>>> be no loss in keeping them out of the kobject tree and
>>> by implication sysfs.  The gain in ease of testing
>>> and code coverage should be significant.
>>>
>>> I.e. people running distributions that make it next to
>>> impossible to boot without sysfs should at be able to
>>> boot a test kernel now.
>>>
>>> Plus no ABIs are harmed with this patch.
>
>>> Signed-off-by: Eric W. Biederman <ebiederm at xmission.com>
>>
>> Duh.
>>
>> Tested-by: Serge Hallyn <serue at us.ibm.com>
>> Acked-by: Serge Hallyn <serue at us.ibm.com>
>
> Oh, this patch is short, clean, and the limitation introduced isn't too
> annoying for testing netns right now.
>
> At least, my proposal provoked some reactions :)

Yes.

> BTW, there's a second limitation with your patch:
> we can't rename the net devices in the additional network namespaces.
>
> In net/core/dev.c, dev_change_name() fails: call to device_rename() return an
> (expected) -EINVAL error.
> Maybe we should add a test on the net to only call it in init_net?

Yes.  Good catch.

Eric