tcp: connect() race with timewait reuse

Its currently possible that several threads issuing a connect() find
the same timewait socket and try to reuse it, leading to list
corruptions.

Condition for bug is that these threads bound their socket on same
address/port of to-be-find timewait socket, and connected to same
target. (SO_REUSEADDR needed)

To fix this problem, we could unhash timewait socket while holding
ehash lock, to make sure lookups/changes will be serialized. Only
first thread finds the timewait socket, other ones find the
established socket and return an EADDRNOTAVAIL error.

This second version takes into account Evgeniy's review and makes sure
inet_twsk_put() is called outside of locked sections.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 10 insertions, 5 deletions

diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
index 00c6a3e..c813e29 100644
--- a/net/ipv6/inet6_hashtables.c
+++ b/net/ipv6/inet6_hashtables.c
@@ -223,6 +223,7 @@ static int __inet6_check_established(struct inet_timewait_death_row *death_row,
 	struct sock *sk2;
 	const struct hlist_nulls_node *node;
 	struct inet_timewait_sock *tw;
+	int twrefcnt = 0;
 
 	spin_lock(lock);
 
@@ -250,19 +251,23 @@ unique:
 	 * in hash table socket with a funny identity. */
 	inet->inet_num = lport;
 	inet->inet_sport = htons(lport);
+	sk->sk_hash = hash;
 	WARN_ON(!sk_unhashed(sk));
 	__sk_nulls_add_node_rcu(sk, &head->chain);
-	sk->sk_hash = hash;
+	if (tw) {
+		twrefcnt = inet_twsk_unhash(tw);
+		NET_INC_STATS_BH(net, LINUX_MIB_TIMEWAITRECYCLED);
+	}
 	spin_unlock(lock);
+	if (twrefcnt)
+		inet_twsk_put(tw);
 	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
 
-	if (twp != NULL) {
+	if (twp) {
 		*twp = tw;
-		NET_INC_STATS_BH(net, LINUX_MIB_TIMEWAITRECYCLED);
-	} else if (tw != NULL) {
+	} else if (tw) {
 		/* Silly. Should hash-dance instead... */
 		inet_twsk_deschedule(tw, death_row);
-		NET_INC_STATS_BH(net, LINUX_MIB_TIMEWAITRECYCLED);
 
 		inet_twsk_put(tw);
 	}