netdev
[Top] [All Lists]

Re: [PATCH][RFC] etherip: Ethernet-in-IPv4 tunneling

To: Lennert Buytenhek <buytenh@xxxxxxxxxxxxxx>
Subject: Re: [PATCH][RFC] etherip: Ethernet-in-IPv4 tunneling
From: Thomas Graf <tgraf@xxxxxxx>
Date: Thu, 13 Jan 2005 02:20:07 +0100
Cc: Ben Greear <greearb@xxxxxxxxxxxxxxx>, netdev@xxxxxxxxxxx, shemminger@xxxxxxxx, shollenbeck@xxxxxxxxxxxx
In-reply-to: <20050113003625.GJ14280@xxxxxxxxxxxxxxxxx>
References: <20050112222437.GC14280@xxxxxxxxxxxxxxxxx> <41E5A7E9.4030101@xxxxxxxxxxxxxxx> <20050112224810.GE14280@xxxxxxxxxxxxxxxxx> <41E5AEAC.8060706@xxxxxxxxxxxxxxx> <20050112231615.GF14280@xxxxxxxxxxxxxxxxx> <20050112234344.GM26856@xxxxxxxxxxxxxx> <20050113001837.GH14280@xxxxxxxxxxxxxxxxx> <20050113002806.GN26856@xxxxxxxxxxxxxx> <20050113003625.GJ14280@xxxxxxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
* Lennert Buytenhek <20050113003625.GJ14280@xxxxxxxxxxxxxxxxx> 2005-01-13 01:36
> On Thu, Jan 13, 2005 at 01:28:06AM +0100, Thomas Graf wrote:
> 
> > > What's that?  I can't find it in my kernel tree nor on google.
> > 
> > Typo, sorry. I meant ip_tunnel_parm. Thinking of it, shouldn't protocol
> > in ip_tunnel_parm->iph->protocol be set to ETHER_IP so userspace could
> > find out this way?
> 
> ip_tunnel_parm->iph->protocol for ether/ip tunnels is IPPROTO_ETHERIP,
> which is 97.  So yeah, the info is in there.  But this doesn't help you
> much in determining whether an arbitrary network device is in fact an
> ether/ip tunnel or not, since SIOCGETTUNNEL aliases with SIOCDEVPRIVATE+3.

Ahh.. You want userspace to be able to tell even if it is unclear
wether it is a tunnel. I think the only way to avoid a new ioctl would
be to introduce IFF_TUNNEL or alike and have userspace call GETTUNNEL
if it is set. This also solves the problem for all other tunnels.
You can also reuse one of the unused flags such as IFF_NOTRAILERS
since it would be read only anyway but I guess this would be too
much of a hack ;->

<Prev in Thread] Current Thread [Next in Thread>