Re: [PATCH] IPv6: fix rt_lookup in pmtu_discovery

["Maciej Żenczykowski" <zenczykowski@gmail.com>]

David, could you comment on this?

2010/1/13 Maciej Żenczykowski <zenczykowski@gmail.com>:
> I believe you are looking for the following patch.
>
> I don't think it solves the problem of source based routing to any
> (significantly) greater extent than the previous patch.
> That would actually require iterating over all IPs assigned to any
> interface on our machine.
>
> However, I really don't understand the code (nor the precise issues
> involved) sufficiently well to be certain of that.
>
> I actually think the ipv4 code path suffers from a very similar problem.
>
> Mind you - I don't think these issues are really significant problems,
> because they just mean we do PMTU once per (dest ip, src ip) instead
> of once per (dest ip).
>
> diff --git a/net/ipv6/route.c b/net/ipv6/route.c
> index c2bd74c..dd8e3b3 100644
> --- a/net/ipv6/route.c
> +++ b/net/ipv6/route.c
> @@ -1562,14 +1562,13 @@ out:
>  *     i.e. Path MTU discovery
>  */
>
> -void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr,
> -                       struct net_device *dev, u32 pmtu)
> +static void rt6_do_pmtu_disc(struct in6_addr *daddr, struct in6_addr *saddr,
> +                            struct net *net, u32 pmtu, int ifindex)
>  {
>        struct rt6_info *rt, *nrt;
> -       struct net *net = dev_net(dev);
>        int allfrag = 0;
>
> -       rt = rt6_lookup(net, daddr, saddr, dev->ifindex, 0);
> +       rt = rt6_lookup(net, daddr, saddr, ifindex, 0);
>        if (rt == NULL)
>                return;
>
> @@ -1637,6 +1636,31 @@ out:
>        dst_release(&rt->u.dst);
>  }
>
> +void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr,
> +                       struct net_device *dev, u32 pmtu)
> +{
> +       struct net *net = dev_net(dev);
> +
> +       /*
> +        * RFC 1981 states that a node "MUST reduce the size of the packets it
> +        * is sending along the path" that caused the Packet Too Big message.
> +        * Since it's not possible in the general case to determine which
> +        * interface was used to send the original packet, we update the MTU
> +        * on the interface that will be used to send future packets. We also
> +        * update the MTU on the interface that received the Packet Too Big in
> +        * case the original packet was forced out that interface with
> +        * SO_BINDTODEVICE or similar. This is the next best thing to the
> +        * correct behaviour, which would be to update the MTU on all
> +        * interfaces.
> +        */
> +       rt6_do_pmtu_disc(daddr, saddr, net, pmtu, 0);
> +       rt6_do_pmtu_disc(daddr, saddr, net, pmtu, dev->ifindex);
> +       /* also support source address based routing */
> +       rt6_do_pmtu_disc(daddr, NULL, net, pmtu, 0);
> +       rt6_do_pmtu_disc(daddr, NULL, net, pmtu, dev->ifindex);
> +}
> +
> +
>  /*
>  *     Misc support functions
>  */
>