rds: fix two RCU related problems
authorCong Wang <xiyou.wangcong@gmail.com>
Tue, 11 Sep 2018 01:27:26 +0000 (18:27 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 26 Sep 2018 06:37:57 +0000 (08:37 +0200)
[ Upstream commit cc4dfb7f70a344f24c1c71e298deea0771dadcb2 ]

When a rds sock is bound, it is inserted into the bind_hash_table
which is protected by RCU. But when releasing rds sock, after it
is removed from this hash table, it is freed immediately without
respecting RCU grace period. This could cause some use-after-free
as reported by syzbot.

Mark the rds sock with SOCK_RCU_FREE before inserting it into the
bind_hash_table, so that it would be always freed after a RCU grace
period.

The other problem is in rds_find_bound(), the rds sock could be
freed in between rhashtable_lookup_fast() and rds_sock_addref(),
so we need to extend RCU read lock protection in rds_find_bound()
to close this race condition.

Reported-and-tested-by: syzbot+8967084bcac563795dc6@syzkaller.appspotmail.com
Reported-by: syzbot+93a5839deb355537440f@syzkaller.appspotmail.com
Cc: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Cc: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Cc: rds-devel@oss.oracle.com
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oarcle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/rds/bind.c

index 5aa3a64..48257d3 100644 (file)
@@ -60,11 +60,13 @@ struct rds_sock *rds_find_bound(__be32 addr, __be16 port)
        u64 key = ((u64)addr << 32) | port;
        struct rds_sock *rs;
 
-       rs = rhashtable_lookup_fast(&bind_hash_table, &key, ht_parms);
+       rcu_read_lock();
+       rs = rhashtable_lookup(&bind_hash_table, &key, ht_parms);
        if (rs && !sock_flag(rds_rs_to_sk(rs), SOCK_DEAD))
                rds_sock_addref(rs);
        else
                rs = NULL;
+       rcu_read_unlock();
 
        rdsdebug("returning rs %p for %pI4:%u\n", rs, &addr,
                ntohs(port));
@@ -157,6 +159,7 @@ int rds_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
                goto out;
        }
 
+       sock_set_flag(sk, SOCK_RCU_FREE);
        ret = rds_add_bound(rs, sin->sin_addr.s_addr, &sin->sin_port);
        if (ret)
                goto out;