diff mbox series

[net,5/5] mptcp: fix dangling connection hang-up

Message ID 20230916-upstream-net-20230915-mptcp-hanging-conn-v1-5-05d1a8b851a8@tessares.net (mailing list archive)
State Accepted
Commit 27e5ccc2d5a50ed61bb73153edb1066104b108b3
Delegated to: Netdev Maintainers
Headers show
Series mptcp: fix stalled connections | expand

Checks

Context Check Description
netdev/series_format success Posting correctly formatted
netdev/tree_selection success Clearly marked for net
netdev/fixes_present success Fixes tag present in non-next series
netdev/header_inline success No static functions without inline keyword in header files
netdev/build_32bit success Errors and warnings before: 1345 this patch: 1345
netdev/cc_maintainers success CCed 9 of 9 maintainers
netdev/build_clang success Errors and warnings before: 1364 this patch: 1364
netdev/verify_signedoff success Signed-off-by tag matches author and committer
netdev/deprecated_api success None detected
netdev/check_selftest success No net selftest shell script
netdev/verify_fixes success Fixes tag looks correct
netdev/build_allmodconfig_warn success Errors and warnings before: 1368 this patch: 1368
netdev/checkpatch warning WARNING: line length of 81 exceeds 80 columns WARNING: line length of 86 exceeds 80 columns WARNING: line length of 91 exceeds 80 columns
netdev/kdoc success Errors and warnings before: 0 this patch: 0
netdev/source_inline success Was 0 now: 0

Commit Message

Matthieu Baerts Sept. 16, 2023, 10:52 a.m. UTC 
From: Paolo Abeni <pabeni@redhat.com>

According to RFC 8684 section 3.3:

  A connection is not closed unless [...] or an implementation-specific
  connection-level send timeout.

Currently the MPTCP protocol does not implement such timeout, and
connection timing-out at the TCP-level never move to close state.

Introduces a catch-up condition at subflow close time to move the
MPTCP socket to close, too.

That additionally allows removing similar existing inside the worker.

Finally, allow some additional timeout for plain ESTABLISHED mptcp
sockets, as the protocol allows creating new subflows even at that
point and making the connection functional again.

This issue is actually present since the beginning, but it is basically
impossible to solve without a long chain of functional pre-requisites
topped by commit bbd49d114d57 ("mptcp: consolidate transition to
TCP_CLOSE in mptcp_do_fastclose()"). When backporting this current
patch, please also backport this other commit as well.

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/430
Fixes: e16163b6e2b7 ("mptcp: refactor shutdown and close")
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Reviewed-by: Mat Martineau <martineau@kernel.org>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
---
 net/mptcp/protocol.c | 86 +++++++++++++++++++++++++---------------------------
 net/mptcp/protocol.h | 22 ++++++++++++++
 net/mptcp/subflow.c  |  1 +
 3 files changed, 65 insertions(+), 44 deletions(-)
diff mbox series

Patch

diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
index c8f38f303a90..e252539b1e19 100644
--- a/net/mptcp/protocol.c
+++ b/net/mptcp/protocol.c
@@ -892,6 +892,7 @@  static bool __mptcp_finish_join(struct mptcp_sock *msk, struct sock *ssk)
 	mptcp_subflow_ctx(ssk)->subflow_id = msk->subflow_id++;
 	mptcp_sockopt_sync_locked(msk, ssk);
 	mptcp_subflow_joined(msk, ssk);
+	mptcp_stop_tout_timer(sk);
 	return true;
 }
 
@@ -2369,18 +2370,14 @@  static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 	bool dispose_it, need_push = false;
 
 	/* If the first subflow moved to a close state before accept, e.g. due
-	 * to an incoming reset, mptcp either:
-	 * - if either the subflow or the msk are dead, destroy the context
-	 *   (the subflow socket is deleted by inet_child_forget) and the msk
-	 * - otherwise do nothing at the moment and take action at accept and/or
-	 *   listener shutdown - user-space must be able to accept() the closed
-	 *   socket.
+	 * to an incoming reset or listener shutdown, the subflow socket is
+	 * already deleted by inet_child_forget() and the mptcp socket can't
+	 * survive too.
 	 */
-	if (msk->in_accept_queue && msk->first == ssk) {
-		if (!sock_flag(sk, SOCK_DEAD) && !sock_flag(ssk, SOCK_DEAD))
-			return;
-
+	if (msk->in_accept_queue && msk->first == ssk &&
+	    (sock_flag(sk, SOCK_DEAD) || sock_flag(ssk, SOCK_DEAD))) {
 		/* ensure later check in mptcp_worker() will dispose the msk */
+		mptcp_set_close_tout(sk, tcp_jiffies32 - (TCP_TIMEWAIT_LEN + 1));
 		sock_set_flag(sk, SOCK_DEAD);
 		lock_sock_nested(ssk, SINGLE_DEPTH_NESTING);
 		mptcp_subflow_drop_ctx(ssk);
@@ -2443,6 +2440,22 @@  static void __mptcp_close_ssk(struct sock *sk, struct sock *ssk,
 out:
 	if (need_push)
 		__mptcp_push_pending(sk, 0);
+
+	/* Catch every 'all subflows closed' scenario, including peers silently
+	 * closing them, e.g. due to timeout.
+	 * For established sockets, allow an additional timeout before closing,
+	 * as the protocol can still create more subflows.
+	 */
+	if (list_is_singular(&msk->conn_list) && msk->first &&
+	    inet_sk_state_load(msk->first) == TCP_CLOSE) {
+		if (sk->sk_state != TCP_ESTABLISHED ||
+		    msk->in_accept_queue || sock_flag(sk, SOCK_DEAD)) {
+			inet_sk_state_store(sk, TCP_CLOSE);
+			mptcp_close_wake_up(sk);
+		} else {
+			mptcp_start_tout_timer(sk);
+		}
+	}
 }
 
 void mptcp_close_ssk(struct sock *sk, struct sock *ssk,
@@ -2486,23 +2499,14 @@  static void __mptcp_close_subflow(struct sock *sk)
 
 }
 
-static bool mptcp_should_close(const struct sock *sk)
+static bool mptcp_close_tout_expired(const struct sock *sk)
 {
-	s32 delta = tcp_jiffies32 - inet_csk(sk)->icsk_mtup.probe_timestamp;
-	struct mptcp_subflow_context *subflow;
+	if (!inet_csk(sk)->icsk_mtup.probe_timestamp ||
+	    sk->sk_state == TCP_CLOSE)
+		return false;
 
-	if (delta >= TCP_TIMEWAIT_LEN || mptcp_sk(sk)->in_accept_queue)
-		return true;
-
-	/* if all subflows are in closed status don't bother with additional
-	 * timeout
-	 */
-	mptcp_for_each_subflow(mptcp_sk(sk), subflow) {
-		if (inet_sk_state_load(mptcp_subflow_tcp_sock(subflow)) !=
-		    TCP_CLOSE)
-			return false;
-	}
-	return true;
+	return time_after32(tcp_jiffies32,
+		  inet_csk(sk)->icsk_mtup.probe_timestamp + TCP_TIMEWAIT_LEN);
 }
 
 static void mptcp_check_fastclose(struct mptcp_sock *msk)
@@ -2641,15 +2645,16 @@  void mptcp_reset_tout_timer(struct mptcp_sock *msk, unsigned long fail_tout)
 	struct sock *sk = (struct sock *)msk;
 	unsigned long timeout, close_timeout;
 
-	if (!fail_tout && !sock_flag(sk, SOCK_DEAD))
+	if (!fail_tout && !inet_csk(sk)->icsk_mtup.probe_timestamp)
 		return;
 
-	close_timeout = inet_csk(sk)->icsk_mtup.probe_timestamp - tcp_jiffies32 + jiffies + TCP_TIMEWAIT_LEN;
+	close_timeout = inet_csk(sk)->icsk_mtup.probe_timestamp - tcp_jiffies32 + jiffies +
+			TCP_TIMEWAIT_LEN;
 
 	/* the close timeout takes precedence on the fail one, and here at least one of
 	 * them is active
 	 */
-	timeout = sock_flag(sk, SOCK_DEAD) ? close_timeout : fail_tout;
+	timeout = inet_csk(sk)->icsk_mtup.probe_timestamp ? close_timeout : fail_tout;
 
 	sk_reset_timer(sk, &sk->sk_timer, timeout);
 }
@@ -2668,8 +2673,6 @@  static void mptcp_mp_fail_no_response(struct mptcp_sock *msk)
 	mptcp_subflow_reset(ssk);
 	WRITE_ONCE(mptcp_subflow_ctx(ssk)->fail_tout, 0);
 	unlock_sock_fast(ssk, slow);
-
-	mptcp_reset_tout_timer(msk, 0);
 }
 
 static void mptcp_do_fastclose(struct sock *sk)
@@ -2706,18 +2709,14 @@  static void mptcp_worker(struct work_struct *work)
 	if (test_and_clear_bit(MPTCP_WORK_CLOSE_SUBFLOW, &msk->flags))
 		__mptcp_close_subflow(sk);
 
-	/* There is no point in keeping around an orphaned sk timedout or
-	 * closed, but we need the msk around to reply to incoming DATA_FIN,
-	 * even if it is orphaned and in FIN_WAIT2 state
-	 */
-	if (sock_flag(sk, SOCK_DEAD)) {
-		if (mptcp_should_close(sk))
-			mptcp_do_fastclose(sk);
+	if (mptcp_close_tout_expired(sk)) {
+		mptcp_do_fastclose(sk);
+		mptcp_close_wake_up(sk);
+	}
 
-		if (sk->sk_state == TCP_CLOSE) {
-			__mptcp_destroy_sock(sk);
-			goto unlock;
-		}
+	if (sock_flag(sk, SOCK_DEAD) && sk->sk_state == TCP_CLOSE) {
+		__mptcp_destroy_sock(sk);
+		goto unlock;
 	}
 
 	if (test_and_clear_bit(MPTCP_WORK_RTX, &msk->flags))
@@ -3016,7 +3015,6 @@  bool __mptcp_close(struct sock *sk, long timeout)
 
 cleanup:
 	/* orphan all the subflows */
-	inet_csk(sk)->icsk_mtup.probe_timestamp = tcp_jiffies32;
 	mptcp_for_each_subflow(msk, subflow) {
 		struct sock *ssk = mptcp_subflow_tcp_sock(subflow);
 		bool slow = lock_sock_fast_nested(ssk);
@@ -3053,7 +3051,7 @@  bool __mptcp_close(struct sock *sk, long timeout)
 		__mptcp_destroy_sock(sk);
 		do_cancel_work = true;
 	} else {
-		mptcp_reset_tout_timer(msk, 0);
+		mptcp_start_tout_timer(sk);
 	}
 
 	return do_cancel_work;
@@ -3117,7 +3115,7 @@  static int mptcp_disconnect(struct sock *sk, int flags)
 	inet_sk_state_store(sk, TCP_CLOSE);
 
 	mptcp_stop_rtx_timer(sk);
-	sk_stop_timer(sk, &sk->sk_timer);
+	mptcp_stop_tout_timer(sk);
 
 	if (msk->token)
 		mptcp_event(MPTCP_EVENT_CLOSED, msk, NULL, GFP_KERNEL);
diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
index 5e2026815c8e..ed61d6850cce 100644
--- a/net/mptcp/protocol.h
+++ b/net/mptcp/protocol.h
@@ -719,6 +719,28 @@  void mptcp_get_options(const struct sk_buff *skb,
 void mptcp_finish_connect(struct sock *sk);
 void __mptcp_set_connected(struct sock *sk);
 void mptcp_reset_tout_timer(struct mptcp_sock *msk, unsigned long fail_tout);
+
+static inline void mptcp_stop_tout_timer(struct sock *sk)
+{
+	if (!inet_csk(sk)->icsk_mtup.probe_timestamp)
+		return;
+
+	sk_stop_timer(sk, &sk->sk_timer);
+	inet_csk(sk)->icsk_mtup.probe_timestamp = 0;
+}
+
+static inline void mptcp_set_close_tout(struct sock *sk, unsigned long tout)
+{
+	/* avoid 0 timestamp, as that means no close timeout */
+	inet_csk(sk)->icsk_mtup.probe_timestamp = tout ? : 1;
+}
+
+static inline void mptcp_start_tout_timer(struct sock *sk)
+{
+	mptcp_set_close_tout(sk, tcp_jiffies32);
+	mptcp_reset_tout_timer(mptcp_sk(sk), 0);
+}
+
 static inline bool mptcp_is_fully_established(struct sock *sk)
 {
 	return inet_sk_state_load(sk) == TCP_ESTABLISHED &&
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index 433f290984c8..918c1a235790 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1552,6 +1552,7 @@  int __mptcp_subflow_connect(struct sock *sk, const struct mptcp_addr_info *loc,
 	mptcp_sock_graft(ssk, sk->sk_socket);
 	iput(SOCK_INODE(sf));
 	WRITE_ONCE(msk->allow_infinite_fallback, false);
+	mptcp_stop_tout_timer(sk);
 	return 0;
 
 failed_unlink: