net: Protect iterations over net::fib_notifier_ops in fib_seq_sum()

There is at least unlocked deletion of net->ipv4.fib_notifier_ops
from net::fib_notifier_ops:

ip_fib_net_exit()
  rtnl_unlock()
  fib4_notifier_exit()
    fib_notifier_ops_unregister(net->ipv4.notifier_ops)
      list_del_rcu(&ops->list)

So fib_seq_sum() can't use rtnl_lock() only for protection.

The possible solution could be to use rtnl_lock()
in fib_notifier_ops_unregister(), but this adds
a possible delay during net namespace creation,
so we better use rcu_read_lock() till someone
really needs the mutex (if that happens).

Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/core/fib_notifier.c b/net/core/fib_notifier.c
index 6b8cd494574fb0908ea0a37659b75486706289c2..0c048bdeb0160f1d7f5865e59390f3824c8a1fe8 100644 (file)
--- a/net/core/fib_notifier.c
+++ b/net/core/fib_notifier.c
@@ -34,12 +34,14 @@ static unsigned int fib_seq_sum(void)
 
        rtnl_lock();
        for_each_net(net) {
-               list_for_each_entry(ops, &net->fib_notifier_ops, list) {
+               rcu_read_lock();
+               list_for_each_entry_rcu(ops, &net->fib_notifier_ops, list) {
                        if (!try_module_get(ops->owner))
                                continue;
                        fib_seq += ops->fib_seq_read(net);
                        module_put(ops->owner);
                }
+               rcu_read_unlock();
        }
        rtnl_unlock();