On Tue, 2020-11-24 at 17:24 +0100, Florian Westphal wrote:
> On close this timer might be scheduled. mptcp uses sk_reset_timer for
> this, so the a reference on the mptcp socket is taken.
>
> This causes a refcount leak which can for example be reproduced
> with 'mp_join_server_v4.pkt' from the mptcp-packetdrill repo.
Whoops, my fault!
> The leak has nothing to do with join requests, v1_mp_capable_bind_no_cs.pkt
> works too when replacing the last ack mpcapable to v1 instead of v0.
>
> unreferenced object 0xffff888109bba040 (size 2744):
> comm "packetdrill", [..]
> backtrace:
> [..] sk_prot_alloc.isra.0+0x2b/0xc0
> [..] sk_clone_lock+0x2f/0x740
> [..] mptcp_sk_clone+0x33/0x1a0
> [..] subflow_syn_recv_sock+0x2b1/0x690 [..]
>
> Fixes: e16163b6e2b7 ("mptcp: refactor shutdown and close")
> Cc: Paolo Abeni <pab...@redhat.com>
> Cc: Davide Caratti <dcara...@redhat.com>
> Signed-off-by: Florian Westphal <f...@strlen.de>
> ---
> net/mptcp/protocol.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index 4b7794835fea..dc979571f561 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -1710,6 +1710,7 @@ static void mptcp_timeout_timer(struct timer_list *t)
> struct sock *sk = from_timer(sk, t, sk_timer);
>
> mptcp_schedule_work(sk);
> + sock_put(sk);
> }
>
> /* Find an idle subflow. Return NULL if there is unacked data at tcp
LGTM, thanks!
Acked-by: Paolo Abeni <pab...@redhat.com>
