Re: Netrom & linuxnode qstn..

From: Tomi Manninen (wsqnmyxr@mailit.tunk.net)
Date: Sat Sep 29 2001 - 18:47:45 EEST

  • Next message: Ken Adams: "Distro"

    On Sat, 29 Sep 2001, pa3gcu wrote:

    > > Nodes with asterisk in the mnemonic field are never propagated in routing
    > > broadcasts. Also that asterisk is not really the mnemonic. It's just
    > > something the kernel shows when then mnemonic field is _empty_. Anyway
    > > those routes also have quality=0 so they are _not_propagated_.
    >
    > Firstly they do get propergated, secondly the nodes with the asterisk given
    > to them by the kernel get repropergated as well and on top of that they all
    > have a quality and certainly not "0".

    Ok. What I said above is slightly wrong. Linux will not propagate routes
    with quality=0. If a netrom frame arrives from an unknown neighbour, linux
    creates a return route entry with empty mnemonic and a quality of 0. Those
    automatically created entries are thus not propagated. If you see such
    broadcasts (originating from you), that's a bug.

    > No i dont see asterisks in broadcasts, what i see is the following.
    >
    > axip: fm XE2EBC-5 to NODES ctl UI^ pid=CF(NET/ROM) len 238
    > NET/ROM Routing: CICESE
    > DB0CRL VE1DRG-7 147
    >
    > I then see with cat /proc/net/nr_nodes | grep -e "*"
    > DB0CRL-0 * 1 1 137 3 00006

    Ok. So XE2EBC-5 claims it knows DB0CRL. He also claims DB0CRL doesn't have
    a mnemonic. What should we do?

    We could ignore the entry but I'm not sure that is really correct. I don't
    think the NET/ROM spec explicitly says every node must have a mnemonic.
    However maybe we should add a switch to netromd for this anyway.

    > > If, on the other hand, you see a broadcast with an empty mnemonic field
    > > then that is something I'd like to know more about. I don't think this is
    > > possible either however.
    >
    > I'm afriad it is.

    I now see how it is possible. It's however not something we can do much
    about. Except that we can ignore broadcasts that propagate those (and in
    effect call the neighbour a "liar").

    > axip: fm W5DAV-7 to PA3GCU-9 ctl SABM+
    > axip: fm PA3GCU-9 to W5DAV-7 ctl UA-
    > axip: fm W5DAV-7 to PA3GCU-9 ctl I00^ pid=CF(NET/ROM) len 238
    > NET/ROM: YO4GAB-2->PA3GCU-9 ttl 66
    > protocol family 43, proto 55
    > axip: fm PA3GCU-9 to W5DAV-7 ctl RR1v
    >
    > Now YO4GAB-2 shows up in my nodelist with an asterisk and a quality of "0"

    These are the automatic return routes I have been talking about. Every
    existing NET/ROM implementation does this. Your station won't propagate
    these forward however.

    Again we could make the kernel ignore these. It would need to be settable
    however as I think it's not quite the right thing to do in all situations.

    > In those packets are hidden nodes, note listen -a does not show the data

    Listen does not show the data as it doesn't know how to do it. Maybe we
    should make listen do an ascii dump of the content here.

    -- 
    Tomi Manninen           Internet:  zjlul.bhizgdyq@rele.tunk.net
    OH2BNS                  AX.25:     zcpp.raczwq@lhsystems.com
    KP20ME04                Amprnet:   mae@4control.de
    

    - To unsubscribe from this list: send the line "unsubscribe linux-hams" in the body of a message to tcft.atvvlv@rele.tunk.net More majordomo info at http://vger.kernel.org/majordomo-info.html



    This archive was generated by hypermail 2b30 : Sat Sep 29 2001 - 18:50:20 EEST