batman-adv: fix reference counting imbalance while sending fragment
authorAntonio Quartulli <antonio@open-mesh.com>
Wed, 23 Apr 2014 12:05:16 +0000 (14:05 +0200)
committerAntonio Quartulli <antonio@meshcoding.com>
Thu, 15 May 2014 18:03:17 +0000 (20:03 +0200)
In the new fragmentation code the batadv_frag_send_packet()
function obtains a reference to the primary_if, but it does
not release it upon return.

This reference imbalance prevents the primary_if (and then
the related netdevice) to be properly released on shut down.

Fix this by releasing the primary_if in batadv_frag_send_packet().

Introduced by ee75ed88879af88558818a5c6609d85f60ff0df4
("batman-adv: Fragment and send skbs larger than mtu")

Cc: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Acked-by: Martin Hundebøll <martin@hundeboll.net>
net/batman-adv/fragmentation.c

index bcc4bea..f14e54a 100644 (file)
@@ -418,12 +418,13 @@ bool batadv_frag_send_packet(struct sk_buff *skb,
                             struct batadv_neigh_node *neigh_node)
 {
        struct batadv_priv *bat_priv;
-       struct batadv_hard_iface *primary_if;
+       struct batadv_hard_iface *primary_if = NULL;
        struct batadv_frag_packet frag_header;
        struct sk_buff *skb_fragment;
        unsigned mtu = neigh_node->if_incoming->net_dev->mtu;
        unsigned header_size = sizeof(frag_header);
        unsigned max_fragment_size, max_packet_size;
+       bool ret = false;
 
        /* To avoid merge and refragmentation at next-hops we never send
         * fragments larger than BATADV_FRAG_MAX_FRAG_SIZE
@@ -483,7 +484,11 @@ bool batadv_frag_send_packet(struct sk_buff *skb,
                           skb->len + ETH_HLEN);
        batadv_send_skb_packet(skb, neigh_node->if_incoming, neigh_node->addr);
 
-       return true;
+       ret = true;
+
 out_err:
-       return false;
+       if (primary_if)
+               batadv_hardif_free_ref(primary_if);
+
+       return ret;
 }