Eric Dumazet [off-list ref] wrote:

On Mon, 2016-11-28 at 15:43 -0800, Daniele Di Proietto wrote:

quoted

If nf_ct_frag6_gather() returns an error other than -EINPROGRESS, it
means that we still have a reference to the skb.  We should free it
before returning from handle_fragments, as stated in the comment above.

Fixes: daaa7d647f81 ("netfilter: ipv6: avoid nf_iterate recursion")
CC: Florian Westphal <fw@strlen.de>
CC: Pravin B Shelar <redacted>
CC: Joe Stringer <redacted>
Signed-off-by: Daniele Di Proietto <redacted>
---
 net/openvswitch/conntrack.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
index 31045ef..fecefa2 100644
--- a/net/openvswitch/conntrack.c
+++ b/net/openvswitch/conntrack.c

@@ -370,8 +370,11 @@ static int handle_fragments(struct net *net, struct sw_flow_key *key,
 		skb_orphan(skb);
 		memset(IP6CB(skb), 0, sizeof(struct inet6_skb_parm));
 		err = nf_ct_frag6_gather(net, skb, user);
-		if (err)
+		if (err) {
+			if (err != -EINPROGRESS)
+				kfree_skb(skb);
 			return err;
+		}
 
 		key->ip.proto = ipv6_hdr(skb)->nexthdr;
 		ovs_cb.mru = IP6CB(skb)->frag_max_size;

Interesting, have you followed the "GPF in eth_header" thread today ?

In a nutshell, we want a complete patch, not something that would solve
part of the problem.

I think this patch is fine, intent seems to be to only take fully reassembled
skb, rather than a stray fragment (ovs does NOT seem to call handle_fragments
in case skb is already known to not contain a fragment header, afaics).

I'll send a patch for the GPF in eth_header thing soon.