Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 21 Apr 1996 09:01:01 +0200 (SAT)
From:      John Hay <jhay@mikom.csir.co.za>
To:        wollman@lcs.mit.edu (Garrett Wollman)
Cc:        hackers@FreeBSD.ORG (FreeBSD-hackers)
Subject:   Re: IPX Netbios packets.
Message-ID:  <199604210701.JAA15888@zibbi.mikom.csir.co.za>
In-Reply-To: <9604202130.AA19649@halloran-eldar.lcs.mit.edu> from "Garrett Wollman" at Apr 20, 96 05:30:20 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> 
> <<On Sat, 20 Apr 1996 21:38:58 +0200 (SAT), John Hay <jhay@mikom.csir.co.za> said:
> 
> > So I am looking for a nice generic way to be able to disable it per device.
> > I thought of flags with ifconfig, but I don't like that idea too much.
> > Another way might be with the sysctl interface, but I'm not sure how to
> > do it for different interface without having to add code to every network
> > interface in the kernel.
> 
> An interface flag is almost certainly the right way to go.  What you
> need to do is to fix all the interface drivers to not abuse IFF_LINK*
> for medium selection (there is already another field in the struct for
> that purpose), and then use one of the newly-freed flags.

Hmmm. It feels like I had this conversation a while back also. :-)
This does not imply acceptance of the job. :-) I just want to figure out what
must change, and how much work it will be. And hopefully what might break.

The normal network drivers will have to use if_physical then I presume?
This means that ifconfig will have to be taught new things. What should
it be called? phys0 - phys32, or something more understandable? Is there any
other user level programs that will have to be changed? What should happen
with slip? It uses all three IFF_LINK flags for different things -
compression, icmp suppression and auto enable compression.

> 
> Or, if there is some sort of ``address'' associated with this service,
> you might define a special sort of address which, when configured on
> the interface, causes the special behavior to happen.
> 

Maybe I should just explain how IPX type 20 packets work.

It is mostly used by Netbios and other protocols that don't understand
the idea of a router. A IPX protocol type 20 packet will be broadcast
by all IPX routers for 8 hops far. It has a field with space for 8 IPX
network addresses and every router add the network address of his
receiving interface there. Before it transmits a packet it must check
that the network address of the destination net is not in the list. That
is how loops are controlled.

Pretty ugly to your bandwidth. That is why you must be able to switch it
of for certain interfaces.

John
-- 
John Hay -- John.Hay@mikom.csir.co.za



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199604210701.JAA15888>