Re: [manet] AODVv2 implementation

[Henning Rogge <[e-mail]>]

Hi Charlie,

yes it would be great to learn how much tracking a route needs... and
maybe we find a good way to do this tracking without using userspace
forwarding, e.g. doing some counting with nftables and/or ebpf-helpers
(at least for Linux).

For detecting a route request for a non-existing route my idea always
was to use a TUN device and point a route to the whole MANET (or even
the default route) into it. This way normal (existing) routes have
priority but you get traffic going somewhere else into userspace to
parse it without needing a raw-socket.

Henning Rogge

On Sat, Feb 3, 2024 at 12:51 AM Charlie Perkins
<[e-mail]> wrote:
>
> Hello Henning,
>
> As I remember, there were some more theoretical reasons why routes
> should be timed out, but I would need to go back a long time to
> resurrect the exact reasoning.  As an error-prone guess, it could have
> to do with sequence number rollover.  Anyway, after a certain amount of
> time a route is considered invalid.  Nevertheless, the route should be
> kept in order to maintain pertinent information information.
>
> We might want to revisit this and determine how strong the mandate
> should be.
>
> Regards,
> Charlie P.
>
>
> On 12/4/2023 1:03 AM, Henning Rogge wrote:
> > Hi,
> >
> > I was thinking about how to implement AODV2 on a TUN but hit an
> > interesting issue when reading through the "Interaction with the
> > Forwarding Plane" chapter of AODVv2...
> >
> > Chapter 6.4:
> > "AODVv2 needs to update the record of when a route was last used to
> > forward a packet".
> >
> > Does this mean AODV cannot use the kernel routing forwarding system
> > and is expected to do everything in user space? I ask because the
> > "routing cache" (and its statistics for each route) were removed from
> > Linux in version 3.6.
> >
> > Henning Rogge
>