On Tue, Jan 24, 2006 at 05:00:30PM +0200, Mike M. Volokhov wrote:
> On Mon, 23 Jan 2006 22:44:29 +0100
> Manuel Bouyer <bouyer@antioche.eu.org> wrote:
> 
> > On Mon, Jan 23, 2006 at 03:09:08PM +0200, Mike M. Volokhov wrote:
> > > This may be beacuse of xennet/bridge tandem. ``The problem occurs when a
> > > new (unlearnt) mac address is seen by bridge which passes to all
> > > if_xennetS. Then all those doms send the packet back to the bridge ttl
> > > - 1 using their src mac, which presumably the bridge learns then.'' See
> > > also private mail from me.
> > 
> > OK, can you try the attached patch (for netbsd-3) ?
> > If should cause xennet and xvif to drop the packet if it's not for their
> > mac address when not in promiscous mode.
> 
> And yet another obscure with this patch. Right now I've lost connection
> with all my single-attached domains (i.e. with only one xennet, dom[134]),
> and dom0 showing the output errors gradually increased on bridge0.

An output error on a bridge can be a failure to allocate memory, or
the send queue of one of the ethernet interfaces is full (in this case,
netstat -id should show some drops). This would mean that the transmit side
of xvif, or receive side of the xennet is stalled.
Does and ifconfig down/up on these interface make things go again ?

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--