Linux-Hams archive - March 1999: Re: FW: Linux IP tunnels - how they work in practice

Re: FW: Linux IP tunnels - how they work in practice

Jonathan NAYLOR (tbp.fygoquqaon@elart.com.pl)
09 Mar 1999 15:30:09 +0100

On 8 Mar 1999, Tomi Manninen wrote:
> > I seem to remember one of the big differences of AXIP to IPIP is the
> > need to have a look- up table available to both the encapsulator and
> > decapsulator, although for the life of me, I can't remember why :-(
>
> Well at least if you were doing it the xNOS way (which IMHO is better than
> what you get with ax25ipd), you need to map the incoming AXIP frame to an
> (virtual?) axip device.

I can't remember what I was going to do with AXIP now. I had certainly planned to
create a virtual device (axip0 ...) and have all incoming AXIP frames mapped onto
this device.

There would need to be at least one lookup table to map AX.25 callsigns to IP
addesses.

My original idea was to have one AXIP device per virtual link, and so the receive
side would need to lookup the table to map incoming frames to the right AXIP device.
I was going to also add validation to reject incoming AXIP frames from non-partner
stataions.

But if you assume:

(a) that you accept all incoming AXIP frames,
(b) that you use one AXIP device for all AXIP connections,

then you don't need to access the AXIP lookup table on receive. The structure of the
AXIP modules would be similar to the IPIP ones, with one for transmit and one for
receive.

As I said, I did write the AXIP code but lost it a while ago. It wasn't finished, and
it wasn't working, the first outgoing frame would lock up the machine :-(

Jonathan HB9/G4KLX