net/mlx5e: Fix nullptr in add_vlan_push_action()
authorDima Chumak <dchumak@nvidia.com>
Mon, 26 Apr 2021 12:16:26 +0000 (15:16 +0300)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 3 Jun 2021 07:00:36 +0000 (09:00 +0200)
commit dca59f4a791960ec73fa15803faa0abe0f92ece2 upstream.

The result of dev_get_by_index_rcu() is not checked for NULL and then
gets dereferenced immediately.

Also, the RCU lock must be held by the caller of dev_get_by_index_rcu(),
which isn't satisfied by the call stack.

Fix by handling nullptr return value when iflink device is not found.
Add RCU locking around dev_get_by_index_rcu() to avoid possible adverse
effects while iterating over the net_device's hlist.

It is safe not to increment reference count of the net_device pointer in
case of a successful lookup, because it's already handled by VLAN code
during VLAN device registration (see register_vlan_dev and
netdev_upper_dev_link).

Fixes: 278748a95aa3 ("net/mlx5e: Offload TC e-switch rules with egress VLAN device")
Addresses-Coverity: ("Dereference null return value")
Signed-off-by: Dima Chumak <dchumak@nvidia.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/net/ethernet/mellanox/mlx5/core/en_tc.c

index 3079a82..e60b8bc 100644 (file)
@@ -4025,8 +4025,12 @@ static int add_vlan_push_action(struct mlx5e_priv *priv,
        if (err)
                return err;
 
-       *out_dev = dev_get_by_index_rcu(dev_net(vlan_dev),
-                                       dev_get_iflink(vlan_dev));
+       rcu_read_lock();
+       *out_dev = dev_get_by_index_rcu(dev_net(vlan_dev), dev_get_iflink(vlan_dev));
+       rcu_read_unlock();
+       if (!*out_dev)
+               return -ENODEV;
+
        if (is_vlan_dev(*out_dev))
                err = add_vlan_push_action(priv, attr, out_dev, action);