mptcp: avoid race on msk state changes

The msk socket state is currently updated in a few spots without
owning the msk socket lock itself.

Some of such operations are safe, as they happens before exposing
the msk socket to user-space and can't race with other changes.

A couple of them, at connect time, can actually race with close()
or shutdown(), leaving breaking the socket state machine.

This change addresses the issue moving such update under the msk
socket lock with the usual:

<acquire spinlock>
<check sk lock onwers>
<ev defer to release_cb>

scheme.

Closes: #56
Fixes: 8fd7380 ("mptcp: fallback in case of simultaneous connect")
Fixes: c3c123d ("net: mptcp: don't hang in mptcp_sendmsg() after TCP fallback")
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net/mptcp/protocol.c

@@ -2896,6 +2896,11 @@ static void mptcp_release_cb(struct sock *sk)
  spin_lock_bh(&sk->sk_lock.slock);
  }
 
+ /* be sure to set the current sk state before tacking actions
+ * depending on sk_state
+ */
+ if (test_and_clear_bit(MPTCP_CONNECTED, &mptcp_sk(sk)->flags))
+ __mptcp_set_connected(sk);
  if (test_and_clear_bit(MPTCP_CLEAN_UNA, &mptcp_sk(sk)->flags))
  __mptcp_clean_una_wakeup(sk);
  if (test_and_clear_bit(MPTCP_ERROR_REPORT, &mptcp_sk(sk)->flags))

net/mptcp/protocol.h

@@ -112,6 +112,7 @@
 #define MPTCP_ERROR_REPORT 8
 #define MPTCP_RETRANSMIT 9
 #define MPTCP_WORK_SYNC_SETSOCKOPT 10
+#define MPTCP_CONNECTED 11
 
 static inline bool before64(__u64 seq1, __u64 seq2)
 {
@@ -600,6 +601,7 @@ void mptcp_get_options(const struct sock *sk,
  struct mptcp_options_received *mp_opt);
 
 void mptcp_finish_connect(struct sock *sk);
+void __mptcp_set_connected(struct sock *sk);
 static inline bool mptcp_is_fully_established(struct sock *sk)
 {
  return inet_sk_state_load(sk) == TCP_ESTABLISHED &&

net/mptcp/subflow.c

@@ -373,6 +373,24 @@ static bool subflow_use_different_dport(struct mptcp_sock *msk, const struct soc
  return inet_sk(sk)->inet_dport != inet_sk((struct sock *)msk)->inet_dport;
 }
 
+void __mptcp_set_connected(struct sock *sk)
+{
+ if (sk->sk_state == TCP_SYN_SENT) {
+ inet_sk_state_store(sk, TCP_ESTABLISHED);
+ sk->sk_state_change(sk);
+ }
+}
+
+static void mptcp_set_connected(struct sock *sk)
+{
+ mptcp_data_lock(sk);
+ if (!sock_owned_by_user(sk))
+ __mptcp_set_connected(sk);
+ else
+ set_bit(MPTCP_CONNECTED, &mptcp_sk(sk)->flags);
+ mptcp_data_unlock(sk);
+}
+
 static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 {
  struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
@@ -381,10 +399,6 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
 
  subflow->icsk_af_ops->sk_rx_dst_set(sk, skb);
 
- if (inet_sk_state_load(parent) == TCP_SYN_SENT) {
- inet_sk_state_store(parent, TCP_ESTABLISHED);
- parent->sk_state_change(parent);
- }
 
  /* be sure no special action on any packet other than syn-ack */
  if (subflow->conn_finished)
@@ -417,6 +431,7 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
  subflow->remote_key);
  MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_MPCAPABLEACTIVEACK);
  mptcp_finish_connect(sk);
+ mptcp_set_connected(parent);
  } else if (subflow->request_join) {
  u8 hmac[SHA256_DIGEST_SIZE];
 
@@ -457,6 +472,7 @@ static void subflow_finish_connect(struct sock *sk, const struct sk_buff *skb)
  } else if (mptcp_check_fallback(sk)) {
 fallback:
  mptcp_rcv_space_init(mptcp_sk(parent), sk);
+ mptcp_set_connected(parent);
  }
  return;
 
@@ -564,6 +580,7 @@ static void mptcp_sock_destruct(struct sock *sk)
 
 static void mptcp_force_close(struct sock *sk)
 {
+ /* the msk is not yet exposed to user-space */
  inet_sk_state_store(sk, TCP_CLOSE);
  sk_common_release(sk);
 }
@@ -1577,10 +1594,7 @@ static void subflow_state_change(struct sock *sk)
  mptcp_rcv_space_init(mptcp_sk(parent), sk);
  pr_fallback(mptcp_sk(parent));
  subflow->conn_finished = 1;
- if (inet_sk_state_load(parent) == TCP_SYN_SENT) {
- inet_sk_state_store(parent, TCP_ESTABLISHED);
- parent->sk_state_change(parent);
- }
+ mptcp_set_connected(parent);
  }
 
  /* as recvmsg() does not acquire the subflow socket for ssk selection