bpf: use skb_to_full_sk helper in bpf_skb_under_cgroup
authorDaniel Borkmann <daniel@iogearbox.net>
Thu, 22 Sep 2016 23:28:35 +0000 (01:28 +0200)
committerDavid S. Miller <davem@davemloft.net>
Fri, 23 Sep 2016 12:40:27 +0000 (08:40 -0400)
We need to use skb_to_full_sk() helper introduced in commit bd5eb35f16a9
("xfrm: take care of request sockets") as otherwise we miss tcp synack
messages, since ownership is on request socket and therefore it would
miss the sk_fullsock() check. Use skb_to_full_sk() as also done similarly
in the bpf_get_cgroup_classid() helper via 2309236c13fe ("cls_cgroup:
get sk_classid only from full sockets") fix to not let this fall through.

Fixes: 4a482f34afcc ("cgroup: bpf: Add bpf_skb_in_cgroup_proto")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/core/filter.c

index 0920c2ac1d009349802d35c40dc4360ccb0ac9d0..e5d997759d5e042d33cc29368baa891ed1e3508c 100644 (file)
@@ -2408,7 +2408,7 @@ BPF_CALL_3(bpf_skb_under_cgroup, struct sk_buff *, skb, struct bpf_map *, map,
        struct cgroup *cgrp;
        struct sock *sk;
 
-       sk = skb->sk;
+       sk = skb_to_full_sk(skb);
        if (!sk || !sk_fullsock(sk))
                return -ENOENT;
        if (unlikely(idx >= array->map.max_entries))