[PATCH] mptcp: fix splat when closing unaccepted socket

[PATCH] mptcp: fix splat when closing unaccepted socket

[Florian Westphal]

From: Paolo Abeni <pabeni@redhat.com>

If userspace exits before calling accept() on a listener that had at least
one new connection ready, we get:

   Attempt to release TCP socket in state 8

This happens because the mptcp socket gets cloned when the TCP connection
is ready, but the socket is never exposed to userspace.

The client additionally sends a DATA_FIN, which brings connection into
CLOSE_WAIT state.  This in turn prevents the orphan+state reset fixup
in mptcp_sock_destruct() from doing its job.

Fixes: 3721b9b64676b ("mptcp: Track received DATA_FIN sequence number and add related helpers")
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/185
Tested-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
---
 net/mptcp/subflow.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 15620bafc544..01f30f8ea710 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -546,8 +546,7 @@ static void mptcp_sock_destruct(struct sock *sk)
 	 * ESTABLISHED state and will not have the SOCK_DEAD flag.
 	 * Both result in warnings from inet_sock_destruct.
 	 */
-
-	if (sk->sk_state == TCP_ESTABLISHED) {
+	if ((1 << sk->sk_state) & (TCPF_ESTABLISHED | TCPF_CLOSE_WAIT)) {
 		sk->sk_state = TCP_CLOSE;
 		WARN_ON_ONCE(sk->sk_socket);
 		sock_orphan(sk);
-- 
2.26.3

Re: [PATCH] mptcp: fix splat when closing unaccepted socket

[Mat Martineau]

On Mon, 3 May 2021, Florian Westphal wrote:

> From: Paolo Abeni <pabeni@redhat.com>
>
> If userspace exits before calling accept() on a listener that had at least
> one new connection ready, we get:
>
>   Attempt to release TCP socket in state 8
>
> This happens because the mptcp socket gets cloned when the TCP connection
> is ready, but the socket is never exposed to userspace.
>
> The client additionally sends a DATA_FIN, which brings connection into
> CLOSE_WAIT state.  This in turn prevents the orphan+state reset fixup
> in mptcp_sock_destruct() from doing its job.
>
> Fixes: 3721b9b64676b ("mptcp: Track received DATA_FIN sequence number and add related helpers")
> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/185
> Tested-by: Florian Westphal <fw@strlen.de>
> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
> ---
> net/mptcp/subflow.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
> index 15620bafc544..01f30f8ea710 100644
> --- a/net/mptcp/subflow.c
> +++ b/net/mptcp/subflow.c
> @@ -546,8 +546,7 @@ static void mptcp_sock_destruct(struct sock *sk)
> 	 * ESTABLISHED state and will not have the SOCK_DEAD flag.
> 	 * Both result in warnings from inet_sock_destruct.
> 	 */
> -
> -	if (sk->sk_state == TCP_ESTABLISHED) {
> +	if ((1 << sk->sk_state) & (TCPF_ESTABLISHED | TCPF_CLOSE_WAIT)) {
> 		sk->sk_state = TCP_CLOSE;
> 		WARN_ON_ONCE(sk->sk_socket);
> 		sock_orphan(sk);
> -- 
> 2.26.3

Thanks Paolo and Florian. Patch looks good - should go to -net right?

Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>


--
Mat Martineau
Intel

Re: [PATCH] mptcp: fix splat when closing unaccepted socket

[Florian Westphal]

Mat Martineau <mathew.j.martineau@linux.intel.com> wrote:
> Thanks Paolo and Florian. Patch looks good - should go to -net right?

Yes, this needs to go to -net.

Thanks!

Re: [PATCH] mptcp: fix splat when closing unaccepted socket

[Matthieu Baerts]

Hi Florian, Paolo, Mat,

On 03/05/2021 12:02, Florian Westphal wrote:
> From: Paolo Abeni <pabeni@redhat.com>
> 
> If userspace exits before calling accept() on a listener that had at least
> one new connection ready, we get:
> 
>    Attempt to release TCP socket in state 8
> 
> This happens because the mptcp socket gets cloned when the TCP connection
> is ready, but the socket is never exposed to userspace.
> 
> The client additionally sends a DATA_FIN, which brings connection into
> CLOSE_WAIT state.  This in turn prevents the orphan+state reset fixup
> in mptcp_sock_destruct() from doing its job.
> 
> Fixes: 3721b9b64676b ("mptcp: Track received DATA_FIN sequence number and add related helpers")
> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/185
> Tested-by: Florian Westphal <fw@strlen.de>
> Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Thank you for the patch and the review.

Now in our tree with Mat's RvB tag:

- eb5185213368: mptcp: fix splat when closing unaccepted socket
- Results: f6df9e9d3336..9c59ba276e75

Builds and tests are now in progress:

https://cirrus-ci.com/github/multipath-tcp/mptcp_net-next/export/20210504T064955
https://github.com/multipath-tcp/mptcp_net-next/actions/workflows/build-validation.yml?query=branch:export/20210504T064955

Cheers,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net