In 
9dbbfb0ab6680c6a85609041011484e6658e7d3c function tipc_sk_reinit
had additional logic added to loop in the event that function
rhashtable_walk_next() returned -EAGAIN. No worries.
However, if rhashtable_walk_start returns -EAGAIN, it does "continue",
and therefore skips the call to rhashtable_walk_stop(). That has
the effect of calling rcu_read_lock() without its paired call to
rcu_read_unlock(). Since rcu_read_lock() may be nested, the problem
may not be apparent for a while, especially since resize events may
be rare. But the comments to rhashtable_walk_start() state:
 * ...Note that we take the RCU lock in all
 * cases including when we return an error.  So you must always call
 * rhashtable_walk_stop to clean up.
This patch replaces the continue with a goto and label to ensure a
matching call to rhashtable_walk_stop().
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
 
        do {
                tsk = ERR_PTR(rhashtable_walk_start(&iter));
-               if (tsk)
-                       continue;
+               if (IS_ERR(tsk))
+                       goto walk_stop;
 
                while ((tsk = rhashtable_walk_next(&iter)) && !IS_ERR(tsk)) {
                        spin_lock_bh(&tsk->sk.sk_lock.slock);
                        msg_set_orignode(msg, tn->own_addr);
                        spin_unlock_bh(&tsk->sk.sk_lock.slock);
                }
-
+walk_stop:
                rhashtable_walk_stop(&iter);
        } while (tsk == ERR_PTR(-EAGAIN));
 }