* FOU RX interface?
@ 2014-10-02 5:14 Andy Lutomirski
2014-10-02 14:44 ` Tom Herbert
0 siblings, 1 reply; 3+ messages in thread
From: Andy Lutomirski @ 2014-10-02 5:14 UTC (permalink / raw)
To: Network Development, Tom Herbert, David S. Miller
Hi-
Sorry for the lack of proper threading here -- I lost the original message.
If I'm understanding the FOU use case correctly, if I set up a FOU
tunnel tun0 that is encapsulated in UDP on eth0, then tun0 packets
will be transmitted on tun0, but incoming packets will show up on eth0
when they're reinjected after stripping the FOU header.
Is this right? I think that, without a way to reinject the received
packets on the tunnel interface, using FOU will be annoying. For
example, writing firewall rules might be tricky. And programs that
use packet sockets or SO_BINDTODEVICE could have a hard time being
configured such that they notice the received packets.
Also, is it even possible to assign a FOU tunnel to a different
network namespace than the device that's being tunneled over? How
will the received packets end up in the right netns?
--Andy
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: FOU RX interface?
2014-10-02 5:14 FOU RX interface? Andy Lutomirski
@ 2014-10-02 14:44 ` Tom Herbert
2014-10-02 15:24 ` Andy Lutomirski
0 siblings, 1 reply; 3+ messages in thread
From: Tom Herbert @ 2014-10-02 14:44 UTC (permalink / raw)
To: Andy Lutomirski; +Cc: Network Development, David S. Miller
On Wed, Oct 1, 2014 at 10:14 PM, Andy Lutomirski <luto@amacapital.net> wrote:
> Hi-
>
> Sorry for the lack of proper threading here -- I lost the original message.
>
> If I'm understanding the FOU use case correctly, if I set up a FOU
> tunnel tun0 that is encapsulated in UDP on eth0, then tun0 packets
> will be transmitted on tun0, but incoming packets will show up on eth0
> when they're reinjected after stripping the FOU header.
>
Incoming FOU packets will still land on the tunnel interface. In FOU
RX the UDP packet is removed and logically re-injected into the
stack-- at this point the packet is IPIP in IP (or sit, GRE) so
appropriate tunnel protocol processing occurs.
> Is this right? I think that, without a way to reinject the received
> packets on the tunnel interface, using FOU will be annoying. For
> example, writing firewall rules might be tricky. And programs that
> use packet sockets or SO_BINDTODEVICE could have a hard time being
> configured such that they notice the received packets.
>
I believe it should work.
> Also, is it even possible to assign a FOU tunnel to a different
> network namespace than the device that's being tunneled over? How
> will the received packets end up in the right netns?
>
Anything you can do with IP tunnels, you should be able to with FOU
enabled IP tunnels. FOU is transparent to IP tunnels on RX.
> --Andy
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: FOU RX interface?
2014-10-02 14:44 ` Tom Herbert
@ 2014-10-02 15:24 ` Andy Lutomirski
0 siblings, 0 replies; 3+ messages in thread
From: Andy Lutomirski @ 2014-10-02 15:24 UTC (permalink / raw)
To: Tom Herbert; +Cc: Network Development, David S. Miller
On Thu, Oct 2, 2014 at 7:44 AM, Tom Herbert <therbert@google.com> wrote:
> On Wed, Oct 1, 2014 at 10:14 PM, Andy Lutomirski <luto@amacapital.net> wrote:
>> Hi-
>>
>> Sorry for the lack of proper threading here -- I lost the original message.
>>
>> If I'm understanding the FOU use case correctly, if I set up a FOU
>> tunnel tun0 that is encapsulated in UDP on eth0, then tun0 packets
>> will be transmitted on tun0, but incoming packets will show up on eth0
>> when they're reinjected after stripping the FOU header.
>>
> Incoming FOU packets will still land on the tunnel interface. In FOU
> RX the UDP packet is removed and logically re-injected into the
> stack-- at this point the packet is IPIP in IP (or sit, GRE) so
> appropriate tunnel protocol processing occurs.
>
Oh, right. That should have been obvious. Thanks for the clarification!
--Andy
>> Is this right? I think that, without a way to reinject the received
>> packets on the tunnel interface, using FOU will be annoying. For
>> example, writing firewall rules might be tricky. And programs that
>> use packet sockets or SO_BINDTODEVICE could have a hard time being
>> configured such that they notice the received packets.
>>
> I believe it should work.
>
>> Also, is it even possible to assign a FOU tunnel to a different
>> network namespace than the device that's being tunneled over? How
>> will the received packets end up in the right netns?
>>
> Anything you can do with IP tunnels, you should be able to with FOU
> enabled IP tunnels. FOU is transparent to IP tunnels on RX.
>
>> --Andy
--
Andy Lutomirski
AMA Capital Management, LLC
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2014-10-02 15:24 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2014-10-02 5:14 FOU RX interface? Andy Lutomirski
2014-10-02 14:44 ` Tom Herbert
2014-10-02 15:24 ` Andy Lutomirski
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.