namespace support requires network modules to say "GPL"

Daniel Lezcano dlezcano at fr.ibm.com
Tue Dec 4 07:19:38 PST 2007


Daniel Lezcano wrote:
> Ben Greear wrote:
>> Eric W. Biederman wrote:
>>> Patrick McHardy <kaber at trash.net> writes:
>>>
>>>  
>>>> Ben Greear wrote:
>>>>   
>>>>> I have a binary module that uses dev_get_by_name...it's sort of a 
>>>>> bridge-like
>>>>> thing and
>>>>> needs user-space to tell it which device to listen for packets on...
>>>>>
>>>>> This code doesn't need or care about name-spaces, so I don't see 
>>>>> how it could
>>>>> really
>>>>> be infringing on the author's code (any worse than loading a binary 
>>>>> driver
>>>>> into the kernel
>>>>> ever does).
>>>>>       
>>>
>>> Regardless of infringement it is incompatible with a complete network
>>> namespace implementation.  Further it sounds like the module you are
>>> describing defines a kernel ABI without being merged and hopes that
>>> ABI will still be supportable in the future.  Honestly I think doing so
>>> is horrible code maintenance policy.
>>>   
>> I don't mind if the ABI changes, so long as I can still use something 
>> similar.
>>
>> The namespace logic is interesting to me in general, but at this point 
>> I can't think of a way that
>> it actually helps this particular module.  All I really need is a way 
>> to grab every frame
>> from eth0 and then transmit it to eth1.  I'm currently doing this by 
>> finding the netdevice
>> and registering a raw-packet protocol (ie, like tcpdump would do).  At 
>> least up to 2.6.23,
>> this does not require any hacks to the kernel and uses only non GPL 
>> exported symbols.
>>
>> Based on my understanding of the namespace logic, if I never add any 
>> namespaces,
>> the general network layout should look similar to how it does today, 
>> so I should have
>> no logical problem with my module.
>>
>>> Once things are largely complete it makes sense to argue with out of
>>> tree module authors that because they don't have network namespace
>>> support in their modules, their modules are broken.     
>> Does this imply that every module that accesses the network code 
>> *must* become
>> GPL simply because it must interact with namespace logic that is 
>> exported as GPL only symbols?
> 
> That's right, with init_net's EXPORT_SYMBOL_GPL and dev_get_xx, we 
> enforce people to be GPL whatever they didn't asked to have the 
> namespaces in their code.
> 
> Eric, why can we simply change EXPORT_SYMBOL_GPL to EXPORT_SYMBOL for 
> init_net ?

Another suggestion/question, is it acceptable to say non-gpl driver 
should use init_task.nsproxy->net_ns instead of &init_net ?

Or does it make sense to have init_net gpl-exported, since we can access 
it through init_task which is exported without gpl mention ?



More information about the Containers mailing list