Re: [PATCH] bonding: L2L3 xmit doesn't support IPv6

[Eric Dumazet <eric.dumazet@gmail.com>]

Le mardi 11 octobre 2011 à 20:39 -0700, Yinglin Sun a écrit :
> On Tue, Oct 11, 2011 at 7:51 PM, Andy Gospodarek <andy@greyhouse.net> wrote:
> >
> >        if (skb->protocol == htons(ETH_P_IP)) {
> > +               struct iphdr *iph = ip_hdr(skb);
> > +               __be16 *layer4hdr = (__be16 *)((u32 *)iph + iph->ihl);
> >                if (!ip_is_fragment(iph) &&
> >                    (iph->protocol == IPPROTO_TCP ||
> >                     iph->protocol == IPPROTO_UDP)) {
> > @@ -3398,7 +3407,18 @@ static int bond_xmit_hash_policy_l34(struct sk_buff *skb, int count)
> >                }
> >                return (layer4_xor ^
> >                        ((ntohl(iph->saddr ^ iph->daddr)) & 0xffff)) % count;
> > -
> > +       } else if (skb->protocol == htons(ETH_P_IPV6)) {
> > +               struct ipv6hdr *ipv6h = ipv6_hdr(skb);
> > +               __be16 *layer4hdrv6 = (__be16 *)((u8 *)ipv6h + sizeof(*ipv6h));
> > +               if (ipv6h->nexthdr == IPPROTO_TCP || ipv6h->nexthdr == IPPROTO_UDP) {
> 
> Does this work if this is a fragmentation packet? and if
> ipv6h->nexthdr is not IPPROTO_TCP/IPPROTO_UDP, it doesn't mean this is
> not TCP/UDP packet. We need to go through the extension header chain
> and look at the last one. It's likely there are some other extension
> headers before L4 header.
> 

Its a best effort.

If first header is not IPPROTO_TCP/UDP, I am not sure its wise to spend
time in hope to find layer4 info (missing anyway in fragments)

By the way I see no bound checking on SKB head : malicious packet could
make bond_xmit_hash_policy_l34() access unitialized memory.

We have same 'fastpath' in __skb_get_rxhash()