Problems with gif tunnels

[Gianmarco Giovannelli]

At 02.10 09/06/2005, Greg 'groggy' Lehey wrote:
 >On Thursday,  9 June 2005 at  1:46:00 +0200, Jeremie Le Hen wrote:
 >> Greg,
 >>
 >>> My understanding is that GRE is to IP as PPP is to SLIP: it allows
 >>> multiple protocols to be encapsulated.  I've done some tracing with
 >>> Ethereal, and the only difference is a four-byte header in front of
 >>> the payload for GRE; in an IP tunnel, it's simply missing.  I've
 >>> written this up in my diary
 >>> (http://www.lemis.com/grog/diary-jun2005.html#8), along with the
 >>> traces.
 >>
 >> yes it's usually a simple four-byte header when doing a simple tunnel.
 >> But from what I have read [1] and according to what Giorgos said,
 >> it seems it can be a lot more longer, depending on the value of the
 >> five first bits of the GRE header.
 >
 >Ah, that seems reasonable.

Hi Greg, I have follow with interest this thread because I had a similar 
problem sometimes ago and we din't succeded in resolve it as I like ...

I had to connect a couple of  a nets with a freebsd box and a linux box 
(not managed by me). They insist to use the ipip tunnel (p:4) and I think I 
should use the  nos-tun interface we had in the base system to let things 
works ourside. But it didn't do the job so we had to switch on an ipsec 
tunnel (esp only) which works quite well except a few things...

Now I see I could simply use the gif interface (which I wrongly suppose did 
only GRE tunnel :-) to connect to an ipip linux tunnel. Is this right ?
And the nos-tun utility is so a basic replacement of the gif interface ?

Thanks ....