mptcp: do not rely on implicit state check in mptcp_listen()
authorPaolo Abeni <pabeni@redhat.com>
Tue, 4 Jul 2023 20:44:34 +0000 (22:44 +0200)
committerDavid S. Miller <davem@davemloft.net>
Wed, 5 Jul 2023 09:51:14 +0000 (10:51 +0100)
commit0226436acf2495cde4b93e7400e5a87305c26054
tree97023eff199cd0b3adaae7eea9d4aa47ec24c505
parent3fffa15bfef48b0ad6424779c03e68ae8ace5acb
mptcp: do not rely on implicit state check in mptcp_listen()

Since the blamed commit, closing the first subflow resets the first
subflow socket state to SS_UNCONNECTED.

The current mptcp listen implementation relies only on such
state to prevent touching not-fully-disconnected sockets.

Incoming mptcp fastclose (or paired endpoint removal) unconditionally
closes the first subflow.

All the above allows an incoming fastclose followed by a listen() call
to successfully race with a blocking recvmsg(), potentially causing the
latter to hit a divide by zero bug in cleanup_rbuf/__tcp_select_window().

Address the issue explicitly checking the msk socket state in
mptcp_listen(). An alternative solution would be moving the first
subflow socket state update into mptcp_disconnect(), but in the long
term the first subflow socket should be removed: better avoid relaying
on it for internal consistency check.

Fixes: b29fcfb54cd7 ("mptcp: full disconnect implementation")
Cc: stable@vger.kernel.org
Reported-by: Christoph Paasch <cpaasch@apple.com>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/414
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/mptcp/protocol.c