[Xastir] dual port TNC

[Tom Russo]

On Tue, Jun 16, 2009 at 09:51:13AM -0700, we recorded a bogon-computron collision of the <archer at eskimo.com> flavor, containing:
> On Tue, 16 Jun 2009, Curt, WE7U wrote:
> 
> > On Tue, 16 Jun 2009, Jason KG4WSV wrote:
> >
> >> An alternative idea was to add a KISS network interface, for use with
> >> things like socat et al.  Would allow the MKISS daemon to talk to
> >> network ports instead of named pipes, and has the added benefit of
> >> allowing xastir to talk to network serial ports.
> >
> > That would be sweet.  I hear there's a possible solution to the
> > remote TNC problem coming out soon from someone else.  I don't know
> > details but I know someone who's beta-testing it.  If it's
> > open-source it'd be a shame for you to duplicate effort along those
> > lines, that's why I'm mentioning it.
> 
> 
> Looked it up, it was in a message to this list from Tom Russo just
> this week:
> 
> > While running Xastir this way is cute, ultimately I'm planning on
> > using the plug mostly as a central TNC server using AX.25
> > networking and a currently-unreleased piece of code called
> > "ldsped" that is a Linux daemon that talks AX.25 networking and
> > the AGWPE protocol (currently in beta, the author provided me with
> > an x86 binary and says he expects to release source in July or
> > August, at which time I plan to try it on the Sheeva).
> 
> So, a slightly different idea from yours.  Says "expects to release
> source" so we can hope it's under a reasonable open-source license.

The author says GPL.  He also says (after a little begging) that he thinks he
might have it on sourceforge in the next few days.  

I have been trying to figure out ways to do this without ldsped, and they
all seem to revolve around "kissnetd" and remserial.

The thing that's missing is a simple program that can tie together two
ax.25 network ports.  Digi_ned's web page describes a way of doing it with 
digi_ned, but since I don't really want to do any digipeating that seems the
wrong approach.

Here's what I've thought might do the trick if that one piece were available:




       +--- ax25bridge---+            2   3   4
       |                 |            pty pty pty
     ax25p1            ax25p2         |   |   |
       |                 |            |   |   |
     KISS TNC           PTY1          |   |   |
                         |            |   |   |
                         +--kissnetd--+---+---+

That is, there's an AX.25 kernel interface to a real tnc, bridged simply
to a second ax.25 interface that talks KISS on a PTY.  kissnetd creates
a fake network of PTYs that all talk KISS to one-another.  The ptys can be
attached to applications, or to remserial.  Obviously, the PTYs are really
pairs of devices, I'm just showing them as one for simplicity of the figure.

Digi_ned's page shows this with digi_ned where I have the bridge.  But
using Digi_ned only as a bridge seems like overkill.

It's that simple bridge that doesn't seem to exist.  It would merely take all 
packets appearing on one AX.25 interface and send them over to the other, in
both directions.  In this way, the three PTYs left free should be a reasonable
sharing of the one real tnc to any application that speaks kiss, and could
be remotable with remserial.

I had briefly thought that one could use kissnetd alone in this trick:


    +---- kissnetd----+----+----+
    |                 |    |    |
   Real TNC          PTY1  PTY2 PTY3
                     |
                   AX251

so that you could share the TNC between two PTYs (attached to remserial), and
still have an AX25 network on the machine for sharing it among local programs.
But kissnetd won't talk to real serial ports, only PTYs.

But perhaps ldsped will be out as soon as its author says and all this will be
unnecessary.  I have tried it out on my linux laptop and it works great --- a 
drop-in replacement for AGWPE for sharing a TNC, and locally run programs 
still get to talk to the TNC through the kernel AX.25 stuff.  I need the 
source, though, to get it running on the SheevaPlug, which is currently 
sitting alone and turned off until I have time to get this working.

-- 
Tom Russo    KM5VY   SAR502   DM64ux          http://www.swcp.com/~russo/
Tijeras, NM  QRPL#1592 K2#398  SOC#236        http://kevan.org/brain.cgi?DDTNM
  In some cultures what I do would be considered normal. 
                                  -- Ineffective daily affirmation