PPP(was: oE) problems with latest -CURRENT

Bjoern A. Zeeb bzeeb-lists at lists.zabbadoz.net
Sun Aug 1 07:07:47 PDT 2004

On Sun, 1 Aug 2004, Maxim Maximov wrote:

> > this is a PPP only problem. I haven't read all the logs from all your
> > mails but if someone could send me a trace file (written with tcpdump
> > -w) it would make it a lot easier to check what's going wrong in LCP
> > negotiation.
> 'tcpdump -w' output attached.
> >
> > Just to make sure. This is NOT a PPPoE problem. PPPoE seems to work
> > pretty fine, else you wouldn't be able to start PPP LCP negotiation.
> >
> Why rp-pppoe works then? PPP is still the same.

of course; the problem had not been visible from the ppp logs. It
looked like everything goes fine apart that the peer doesn't ACK
your CONREQ and thus you go into another negotiation loop.

The problem actually is that the peer will never see your CONFREQ
because it goes like this:

3c 13			Ethertype (NONSTANDARD, 3Com)
11 00 38 dd 00 1a	PPPoE
c0 21			PPP LCP

but it should go like this:

88 64			Ethertype PPPoE
11 00 3e dd 00 15	PPPoE
c0 21			PPP LCP

unles you
	set pppoe 3Com
or changed the sysctl (I think there still is one).

So it seems that either ether.c (command.c, physical.c) change
or the ng_pppoe still have problems selecting the correct type
and default to standard. I am checking the sources to find the place
if glebius isn't faster ;-)

Could you please mail to output of
	sysctl net.graph.nonstandard_pppoe

Bjoern A. Zeeb				bzeeb at Zabbadoz dot NeT

More information about the freebsd-current mailing list