diff options
author | Paolo Abeni <pabeni@redhat.com> | 2023-06-20 18:24:23 +0200 |
---|---|---|
committer | Jakub Kicinski <kuba@kernel.org> | 2023-06-22 07:44:54 +0200 |
commit | 57fc0f1ceaa4016354cf6f88533e20b56190e41a (patch) | |
tree | fb1230e45df2a73eb6443f9c89793f4c6fb10d3c /net/mptcp/pm_netlink.c | |
parent | mptcp: drop legacy code around RX EOF (diff) | |
download | linux-57fc0f1ceaa4016354cf6f88533e20b56190e41a.tar.xz linux-57fc0f1ceaa4016354cf6f88533e20b56190e41a.zip |
mptcp: ensure listener is unhashed before updating the sk status
The MPTCP protocol access the listener subflow in a lockless
manner in a couple of places (poll, diag). That works only if
the msk itself leaves the listener status only after that the
subflow itself has been closed/disconnected. Otherwise we risk
deadlock in diag, as reported by Christoph.
Address the issue ensuring that the first subflow (the listener
one) is always disconnected before updating the msk socket status.
Reported-by: Christoph Paasch <cpaasch@apple.com>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/407
Fixes: b29fcfb54cd7 ("mptcp: full disconnect implementation")
Cc: stable@vger.kernel.org
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: Jakub Kicinski <kuba@kernel.org>
Diffstat (limited to 'net/mptcp/pm_netlink.c')
-rw-r--r-- | net/mptcp/pm_netlink.c | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c index 59f8f3124855..1224dfca5bf3 100644 --- a/net/mptcp/pm_netlink.c +++ b/net/mptcp/pm_netlink.c @@ -1047,6 +1047,7 @@ static int mptcp_pm_nl_create_listen_socket(struct sock *sk, if (err) return err; + inet_sk_state_store(newsk, TCP_LISTEN); err = kernel_listen(ssock, backlog); if (err) return err; |