* Stefan Berger (stefanb(a)us.ibm.com) wrote:
Daniel Veillard <veillard(a)redhat.com> wrote on 05/11/2010
06:07:58 AM:
>
> On Mon, May 10, 2010 at 07:57:37PM -0400, Stefan Berger wrote:
> > Below is David Alan's original patch with lots of changes.
> >
> > In particular, it now parses the following XML and stored the data
> > internally. No sending of netlink messages has been implemented here.
> >
> > <interface type='direct'>
> > <source dev='static' mode='vepa'/>
> > <model type='virtio'/>
> > <vsi managerid='12' typeid='0x123456'
typeidversion='1'
> > instanceid='fa9b7fff-b0a0-4893-8e0e-beef4ff18f8f' />
> > <filterref filter='clean-traffic'/>
> > </interface>
> >
> > <interface type='direct'>
> > <source dev='static' mode='vepa'/>
> > <model type='virtio'/>
> > <vsi profileid='my_profile'/>
> > </interface>
> >
> > I'd suggest to use this patch as a base for sending out netlink
> > messages.
> >
> > Signed-off-by: Stefan Berger <stefanb(a)us.ibm.com>
>
> I have 2 basic questions:
>
> - how can we be sure that if we apply this the XML interface won't
> need to change due to kernel interfaces, maybe Chris can confirm
> that the kernel is now stable enough that this should not affect
> the XML anymore
I have been trying to gather the parameters necessary to trigger the
slightly
different setup protocols. I cannot be sure that this XML won't change at
the
moment but would also wait to have it checked in. However, the parser here
works to the best of my current knowledge.
The netlink messages that require these parameters drive the requirements
to the XML. So, I'd wait for the kernel implementation to become stable.
> - can I get some signoff from Cisco that there is agreement on this
> format and that will be sufficient for their need
I'd also like to have 100% re-assurance that the TWO pre-standards we are
dealing
with here are not going change later on and cause changes affecting the
XML.
Honestly, the only way you'll get that is to wait until they are
standards (and even then there's possible changes, additions, etc).
The low-level TLVs and the protocol ladder diagrams are not under
substantial churn. Names change regularly, but the notion of
preassociate/associate/deassociate and the basic id's for VDP payload
(database id, profile id, mac+vlan pairs, etc).
thanks,
-chris