[Devel] 2.6.26 panic (skb->dev==NULL), NFS support

[Denis V. Lunev]

On Fri, 2008-08-01 at 10:12 +0400, Denis V. Lunev wrote:
> On Fri, 2008-08-01 at 02:26 +0200, Karel Tuma wrote:
> > Hello list,
> > 
> > I hope this is the right place to post.
> > 
> > I've recently moved to 2.6.26 git (i hope that's the bleeding edge) OpenVZ
> > to find out how usable it is. I'm running it on a box under fair IO load
> > (100-300 BIO/s). The thing panics in net/ipv4/tcp_ipv4.c:tcp_v4_send_ack()
> > once every couple of hours, during io spikes. Apparently skb->dev is NULL,
> > inside an irq context on top of that. It's very hard to pinpoint the actual
> > trigger. Vanilla 2.6.26 runs just fine.
> > 
> > Treating the symptoms, rather than the reason behind it, by testing
> > the pointer seems to work ok, without disrupting any service. However,
> > it would be nice to figure out the real cause to avoid such a hideous hack.
> > 
> > NFS seems to work (mounting from CT0 so far) after minor changes for
> > /proc interfacing (it was throwing ENOMEM). It's mounted in udp so it's
> > unlikely to be the cause for tcp panic.
> > 
> > oops, ugly fix and nfs patches are attached.
> 
> good place for this stuff is bugzilla.openvz.org :) Could you describe a
> bit your setup: what modules are loaded, how network is configured,
> which .config is used?

oops. I forget one more thing :) Could you instead of printing weird message
print skb content including IP from, IP to, protocol (tcp/udp/whatever)
and other stuff?

This could help us with guessing how the packet comes to us. Whether it
is NAT one, multicast, etc. I have no ideas right now.

> Unfortunately, NFS right now is not VE aware in 2.6.26 entirely :( I am
> working on this problem and hope something mainstreamable will be ready
> quite soon.
> 
> Regards,
> 	Den