[PATCH nf] netfilter: ipv6: fix oops when defragmenting locally generated fragments

[PATCH nf] netfilter: ipv6: fix oops when defragmenting locally generated fragments

[Florian Westphal]

Unlike ipv4 and normal ipv6 defrag, netfilter ipv6 defragmentation did
not save/restore skb->dst.

This causes oops when handling locally generated ipv6 fragments, as
output path needs a valid dst.

Reported-by: Maciej Żenczykowski <zenczykowski@gmail.com>
Fixes: 84379c9afe01 ("netfilter: ipv6: nf_defrag: drop skb dst before queueing")
Signed-off-by: Florian Westphal <fw@strlen.de>
---
 net/ipv6/netfilter/nf_conntrack_reasm.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
index 8f68a518d9db..f76bd4d15704 100644
--- a/net/ipv6/netfilter/nf_conntrack_reasm.c
+++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
@@ -587,11 +587,16 @@ int nf_ct_frag6_gather(struct net *net, struct sk_buff *skb, u32 user)
 	 */
 	ret = -EINPROGRESS;
 	if (fq->q.flags == (INET_FRAG_FIRST_IN | INET_FRAG_LAST_IN) &&
-	    fq->q.meat == fq->q.len &&
-	    nf_ct_frag6_reasm(fq, skb, dev))
-		ret = 0;
-	else
+	    fq->q.meat == fq->q.len) {
+		unsigned long orefdst = skb->_skb_refdst;
+
+		skb->_skb_refdst = 0UL;
+		if (nf_ct_frag6_reasm(fq, skb, dev))
+			ret = 0;
+		skb->_skb_refdst = orefdst;
+	} else {
 		skb_dst_drop(skb);
+	}
 
 out_unlock:
 	spin_unlock_bh(&fq->q.lock);
-- 
2.18.1

Re: [PATCH nf] netfilter: ipv6: fix oops when defragmenting locally generated fragments

[Eric Dumazet]

On Tue, Oct 23, 2018 at 7:48 AM Florian Westphal <fw@strlen.de> wrote:
>
> Unlike ipv4 and normal ipv6 defrag, netfilter ipv6 defragmentation did
> not save/restore skb->dst.
>
> This causes oops when handling locally generated ipv6 fragments, as
> output path needs a valid dst.
>
> Reported-by: Maciej Żenczykowski <zenczykowski@gmail.com>
> Fixes: 84379c9afe01 ("netfilter: ipv6: nf_defrag: drop skb dst before queueing")
> Signed-off-by: Florian Westphal <fw@strlen.de>

Reviewed-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH nf] netfilter: ipv6: fix oops when defragmenting locally generated fragments

[Maciej Żenczykowski]

On Tue, Oct 23, 2018 at 7:47 AM, Florian Westphal <fw@strlen.de> wrote:
> Unlike ipv4 and normal ipv6 defrag, netfilter ipv6 defragmentation did
> not save/restore skb->dst.
>
> This causes oops when handling locally generated ipv6 fragments, as
> output path needs a valid dst.
>
> Reported-by: Maciej Żenczykowski <zenczykowski@gmail.com>
> Fixes: 84379c9afe01 ("netfilter: ipv6: nf_defrag: drop skb dst before queueing")
> Signed-off-by: Florian Westphal <fw@strlen.de>
> ---
>  net/ipv6/netfilter/nf_conntrack_reasm.c | 13 +++++++++----
>  1 file changed, 9 insertions(+), 4 deletions(-)
>
> diff --git a/net/ipv6/netfilter/nf_conntrack_reasm.c b/net/ipv6/netfilter/nf_conntrack_reasm.c
> index 8f68a518d9db..f76bd4d15704 100644
> --- a/net/ipv6/netfilter/nf_conntrack_reasm.c
> +++ b/net/ipv6/netfilter/nf_conntrack_reasm.c
> @@ -587,11 +587,16 @@ int nf_ct_frag6_gather(struct net *net, struct sk_buff *skb, u32 user)
>          */
>         ret = -EINPROGRESS;
>         if (fq->q.flags == (INET_FRAG_FIRST_IN | INET_FRAG_LAST_IN) &&
> -           fq->q.meat == fq->q.len &&
> -           nf_ct_frag6_reasm(fq, skb, dev))
> -               ret = 0;
> -       else
> +           fq->q.meat == fq->q.len) {
> +               unsigned long orefdst = skb->_skb_refdst;
> +
> +               skb->_skb_refdst = 0UL;
> +               if (nf_ct_frag6_reasm(fq, skb, dev))
> +                       ret = 0;
> +               skb->_skb_refdst = orefdst;
> +       } else {
>                 skb_dst_drop(skb);
> +       }
>
>  out_unlock:
>         spin_unlock_bh(&fq->q.lock);
> --
> 2.18.1
>

I don't quite follow how this fixes things, but I'll trust you on it.
(nor do I understand why only 4.9 LTS appears to crash with a null ptr deref)

Thanks for the fix.

Re: [PATCH nf] netfilter: ipv6: fix oops when defragmenting locally generated fragments

[Florian Westphal]

Maciej Żenczykowski <maze@google.com> wrote:
> >         ret = -EINPROGRESS;
> >         if (fq->q.flags == (INET_FRAG_FIRST_IN | INET_FRAG_LAST_IN) &&
> > -           fq->q.meat == fq->q.len &&
> > -           nf_ct_frag6_reasm(fq, skb, dev))
> > -               ret = 0;
> > -       else
> > +           fq->q.meat == fq->q.len) {
> > +               unsigned long orefdst = skb->_skb_refdst;
> > +
> > +               skb->_skb_refdst = 0UL;
> > +               if (nf_ct_frag6_reasm(fq, skb, dev))
> > +                       ret = 0;
> > +               skb->_skb_refdst = orefdst;
> > +       } else {
> >                 skb_dst_drop(skb);
> > +       }
> >
> >  out_unlock:
> >         spin_unlock_bh(&fq->q.lock);
> > --
> > 2.18.1
> >
> 
> I don't quite follow how this fixes things, but I'll trust you on it.

The problematic spot is skb_morph() in nf_ct_frag6_reasm(), when we hit
this code path we take dst from a dst-less skb that got queued earlier.

> (nor do I understand why only 4.9 LTS appears to crash with a null ptr deref)

Newer kernels need nft or iptables rule that enables defrag, such as
"ip6tables -A INPUT -m conntrack --ctstate NEW"; 4.9 still enables it by
default.

Re: [PATCH nf] netfilter: ipv6: fix oops when defragmenting locally generated fragments

[Pablo Neira Ayuso]

On Tue, Oct 23, 2018 at 04:47:16PM +0200, Florian Westphal wrote:
> Unlike ipv4 and normal ipv6 defrag, netfilter ipv6 defragmentation did
> not save/restore skb->dst.
> 
> This causes oops when handling locally generated ipv6 fragments, as
> output path needs a valid dst.

Applied, thanks!