bridge: vlan: fix possible null vlgrp deref while registering new port
authorNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Wed, 30 Sep 2015 18:16:52 +0000 (20:16 +0200)
committerDavid S. Miller <davem@davemloft.net>
Fri, 2 Oct 2015 01:24:04 +0000 (18:24 -0700)
While a new port is being initialized the rx_handler gets set, but the
vlans get initialized later in br_add_if() and in that window if we
receive a frame with a link-local address we can try to dereference
p->vlgrp in:
br_handle_frame() -> br_handle_local_finish() -> br_should_learn()

Fix this by checking vlgrp before using it.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/bridge/br_vlan.c

index 283d012c3d89085c62f0e43d80a67d150f971930..678d5c41b55126ae8e99608cdd1206c4e16a5f40 100644 (file)
@@ -476,13 +476,15 @@ bool br_allowed_egress(struct net_bridge_vlan_group *vg,
 /* Called under RCU */
 bool br_should_learn(struct net_bridge_port *p, struct sk_buff *skb, u16 *vid)
 {
+       struct net_bridge_vlan_group *vg;
        struct net_bridge *br = p->br;
 
        /* If filtering was disabled at input, let it pass. */
        if (!br->vlan_enabled)
                return true;
 
-       if (!p->vlgrp->num_vlans)
+       vg = p->vlgrp;
+       if (!vg || !vg->num_vlans)
                return false;
 
        if (!br_vlan_get_tag(skb, vid) && skb->vlan_proto != br->vlan_proto)