Re: The 802.15.4 Security Layer

[Simon Vincent <simon.vincent@xsilon.com>]

Ok yes this makes sense.

Let me know if there is any bits I can help on.

Simon

On 24/06/15 15:01, Alexander Aring wrote:
> On Wed, Jun 24, 2015 at 12:00:14PM +0200, Alexander Aring wrote:
>> Hi,
>>
>> On Tue, Jun 23, 2015 at 04:34:42PM +0100, Simon Vincent wrote:
>>> Hi Alex,
>>>
>>> Do you have your security/nl802154 work available anywhere I can have a
>>> look?
>>> I am in the process of getting 802.15.4 security working on our devices. I
>>> don't want to implement it using the old interface as I will then have to
>>> recode our application when llsec moves over to nl802154.
>> I think what we do at first is a 1:1 implementation of the old interface
>> and the new interface and then look what we can change afterwards if
>> needed.
>>
>> We introduce then some CONFIG_NL802154_EXPERIMENTAL to change the enum
>> definition (with security and without). I think then we are somehow
>> safe to still change the netlink interface, inside kernelspace, afterwards.
>>
> What I meant here is something like [0]. We simple let the config add
> the end of all declaration, if we add something to mainline then we put
> it out of the #ifdef foo. (Above the comments) If we do that then the
> CONFIG_NL802154_EXPERIMENTAL will be broken afterwards and the userspace
> tool need to be updated. Without CONFIG_NL802154_EXPERIMENTAL it should
> always be the same. Just the NL802154_CMD_MAX and NL802154_ATTR_MAX
> will be a lesser value than without CONFIG_NL802154_EXPERIMENTAL. The
> internal nl802154 framework should not react on these definitions then,
> if somebody tries to use CMD's/ATTR's which are inside
> CONFIG_NL802154_EXPERIMENTAL.
>
>
> With CONFIG_NL802154_EXPERIMENTAL then the user need to care about to user
> the right userspace nl802154 header in their application.
>
> - Alex
>
> [0] https://github.com/linux-wpan/linux-wpan-next/commit/4522252b9964227d1a3ce0b09c1aa0a6d95c3ba1