LVS
lvs-devel
Google
 
Web LinuxVirtualServer.org

Re: [PATCH 2/2] ipvs: Extend MTU check to account for IPv6 NAT defrag ch

To: Jesper Dangaard Brouer <brouer@xxxxxxxxxx>
Subject: Re: [PATCH 2/2] ipvs: Extend MTU check to account for IPv6 NAT defrag changes
Cc: netdev@xxxxxxxxxxxxxxx, Patrick McHardy <kaber@xxxxxxxxx>, lvs-devel@xxxxxxxxxxxxxxx, Simon Horman <horms@xxxxxxxxxxxx>, Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx>, Hans Schillstrom <hans@xxxxxxxxxxxxxxx>, Wensong Zhang <wensong@xxxxxxxxxxxx>, netfilter-devel@xxxxxxxxxxxxxxx
From: Julian Anastasov <ja@xxxxxx>
Date: Tue, 28 Aug 2012 12:47:40 +0300 (EEST)
        Hello,

On Tue, 28 Aug 2012, Jesper Dangaard Brouer wrote:

> On Mon, 2012-08-27 at 18:20 +0300, Julian Anastasov wrote:
> > 
> > > @@ -956,8 +963,11 @@ ip_vs_tunnel_xmit_v6(struct sk_buff *skb, struct 
> > > ip_vs_conn *cp,
> > >           skb_dst(skb)->ops->update_pmtu(skb_dst(skb), NULL, skb, mtu);
> > >  
> > >   /* MTU checking: Special for tunnel mode */
> > > - if (mtu < ntohs(old_iph->payload_len) + sizeof(struct ipv6hdr) &&
> 
> I guess:
>  ntohs(old_iph->payload_len) + sizeof(struct ipv6hdr)
> Is the same as:
>  skb->len

        I think so. You can think of this in different way:
all transmitters are called from same place, there is no
difference in the packets we see. When we can use
__mtu_check_toobig_v6 for other methods relying on skb->len
being correct, we can do the same for tunnels, only that
tunnels have lower MTU, that is the only difference.

> > > -     !skb_is_gso(skb)) {
> > > + if ((!IP6CB(skb)->frag_max_size &&
> > > +      (mtu < ntohs(old_iph->payload_len) + sizeof(struct ipv6hdr) &&
> > > +       !skb_is_gso(skb)))
> > > +     || IP6CB(skb)->frag_max_size + sizeof(struct ipv6hdr) > mtu) {
> 
> > 
> >     mtu is already reduced with the new outer header size,
> > may be we can just call __mtu_check_toobig_v6 with mtu?
> 
> To Julian, is the extra sizeof(struct ipv6hdr) addition to
> frag_max_size, wrong? (as the mtu is already reduced)

        Yes, sizeof(struct ipv6hdr) is needed only together
with payload_len because payload_len does not include the
first header.

> If above statements hold, I think we can simply use
> __mtu_check_toobig_v6() also for the tunnel case :-)

        Yep

> --Jesper

Regards

--
Julian Anastasov <ja@xxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe lvs-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

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