sctp: also point GSO head_skb to the sk when it's available

The head skb for GSO packets won't travel through the inner depths of
SCTP stack as it doesn't contain any chunks on it. That means skb->sk
doesn't get set and then when sctp_recvmsg() calls
sctp_inet6_skb_msgname() on the head_skb it panics, as this last needs
to check flags at the socket (sp->v4mapped).

The fix is to initialize skb->sk for th head skb once we are able to do
it. That is, when the first chunk is processed.

Signed-off-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 3 insertions, 0 deletions

diff --git a/net/sctp/ulpevent.c b/net/sctp/ulpevent.c
index f6219b164b42..1bc4f71aaba8 100644
--- a/net/sctp/ulpevent.c
+++ b/net/sctp/ulpevent.c
@@ -91,6 +91,7 @@ int sctp_ulpevent_is_notification(const struct sctp_ulpevent *event)
 static inline void sctp_ulpevent_set_owner(struct sctp_ulpevent *event,
                                            const struct sctp_association *asoc)
 {
+        struct sctp_chunk *chunk = event->chunk;
         struct sk_buff *skb;
 
         /* Cast away the const, as we are just wanting to
@@ -101,6 +102,8 @@ static inline void sctp_ulpevent_set_owner(struct sctp_ulpevent *event,
         event->asoc = (struct sctp_association *)asoc;
         atomic_add(event->rmem_len, &event->asoc->rmem_alloc);
         sctp_skb_set_owner_r(skb, asoc->base.sk);
+        if (chunk && chunk->head_skb && !chunk->head_skb->sk)
+                chunk->head_skb->sk = asoc->base.sk;
 }
 
 /* A simple destructor to give up the reference to the association. */