Linux-Hams archive - January 1997: Re: listen problems.

Re: listen problems.

Tomi Manninen (jwubj@net2000.com.au)
Fri, 3 Jan 1997 17:01:01 +0200 (EET)


On Fri, 3 Jan 1997 qyj@gofilnet.com wrote:

> Richard Adams typed:
> > [Wed Jan 1 15:18:02 1997]
> > Port scc3: EAX25: [invalid]->[invalid] <I S4 R0> pid=0x45 [DAMA]
> Hmm, a pid of 0x45. The PID byte is just before the data payload and the
> first byte in the payload is the IP version and header length. So an ipv4
> packet with header length of 20 would be 0x45 (or 0x54 never get that
> byteorder right!) coincindence?

When this problem was first discussed on this list something like a year
ago i remember Alan Cox saying something like "Yes it's an old problem in
the PI driver, it seems to slip a byte (or was it bit?) in the beginning
of the frame sometimes." That would match your conclusion.

This problem was the reason why I added the callsign validity check in the
listen code. I was annoyed because every now and then a garbled callsign
with ascii 7 (BELL character) caused my computer to beep. And I rushed to
see who sent me mail...

> I have a PT or PackeTwin card. The device drivers for both a very similar.
> I remember seeing some dramitic warnings in the packetwin driver code for
> NOS, perhaps it is time to re-read them again. I would be nice if Don was
> active.

Wasn't the PT driver modified from the PI driver?

> For me and the rest of the PackeTwin users (perhaps PI too) it is a little
> simpler; we've had this problem since the start of the general release of
> the PT code. I cannot remember how it was with the non-GPL code, which used
> a very different system. Perhaps one of the other alpha testers (Ken?)
> could help.

The PI driver has had this problem at least as long as I have used my PI2
card with Linux.

-- 
---Tomi Manninen   /   xmgzr.eegfhall@urban.ne.jp   /   OH2BNS @ OH2RBI.FIN.EU---