----- Original Message -----
From: "Lennert Buytenhek" <[EMAIL PROTECTED]>
To: "Bart De Schuymer" <[EMAIL PROTECTED]>
Cc: "Eric Low" <[EMAIL PROTECTED]>; <[EMAIL PROTECTED]>
Sent: Friday, March 01, 2002 9:41 PM
Subject: Re: [Bridge] locally generated packets lost after output chain


>
> On Fri, Mar 01, 2002 at 09:23:20PM +0100, Bart De Schuymer wrote:
>
> > Try this patch and plz say if it solves it. I posted it some time ago
> > (January 15th), but Lennert ignored/missed it.
>
> I didn't immediately see why it would be necessary, so it got swamped
> under a ton of other mail.  Mea maxima culpa :(

> > If your packets are routed then the bridge-nf patch will make the
packets go
> > through the ebtables FORWARD chain instead of the ebtables OUTPUT chain
> > (without this patch). I don't like it...
>
> If packets are routed, they should not go through br_nf_local_out,
> right?

You can have the situation where the bridge is also a router and e.g. this
setup:
A bridge with on the one side netmask 172.16.1.0 and on the other netmask
172.16.2.0.
If a computer with ip 172.16.1.200 wants to talk to computer with ip
172.16.2.200 it will send the packets to its default gateway, being the
bridge/router. A strange setup I guess, but still.
So: the packet arrives on the bridge hook and is passed up for routing,
later the packet is given back to the same bridge in the LOCAL_OUT hook.
I'm pretty sure back then I tested this (strange) setup.
I'm guessing Eric has some kind of similar situation.

You just make me realize that my patch makes the layer 2 flow (seen from
ebtables' standpoint) for ip DNATed 'bridged' packets unnatural. So we need
a compromise that handles both, right?

cheers,
Bart

_______________________________________________
Bridge mailing list
[EMAIL PROTECTED]
http://www.math.leidenuniv.nl/mailman/listinfo/bridge

Reply via email to