[DNA] RE: Review of link-information from Pekka Savola

[Alper Yegin <alper.yegin@samsung.com>]

Pekka,

I perceive this mostly as a stack configuration issue. If your IP has
nothing to do with a given interface, it should not be getting any
notifications from there (include the link up). [For example, you would
not plumb IP driver on a non-IP L2 driver]. 

IPv4 vs. IPv6 distinction is harder. Link up would tell "IP is enabled",
but not necessarily which version of the IP is supported on that link.
The DNA process that consumes the link up notification would have to
figure that out.

Alper



> >>      Node's establishment of a link-layer connection with an
> > attachment
> >>      point that signifies the availability of IP service (i.e.,
being
> >> able
> >>      to send and receive IP packets) between the two is considered
a
> >> link
> >>      up event.
> >>
> >>  ==> a physical link is a link which can carry any media.  Maybe
> >> you should
> >>  reword this slightly -- for example, a host does not know about
> >> availability
> >>  of IP (or anything else).  Maybe "availability of IP service" is
> >> specific to
> >>  some link types which can only carry IP service, not any L3
protocol?
> >
> > Not sure if I understood this. Could you please elaborate a bit
more?
> 
> You're making the assumption that once a lower-layer link is up, the
> link can be used to transport IP packets.
> 
> What if the link supported only IPX or CLNS packets?  What if the link
> supported only IPv4 but not IPv6 packets? As an example of these,
> consider protocol-based Ethernet VLANs.
> 
> That said, I'm not sure how much this spec can do about this kind of
> situation.  Maybe it could be a bit clearer that it's making a leap of
> faith when associating "link up => IP protocol available" unless the
> link can be known a priori to support IP protocol (e.g., I'd guess if
> you do PPP and also IPCP, you can be pretty sure of that..).