vlan: fix a race in egress prio management
authorEric Dumazet <edumazet@google.com>
Thu, 18 Jul 2013 16:35:10 +0000 (09:35 -0700)
committerDavid S. Miller <davem@davemloft.net>
Thu, 18 Jul 2013 20:07:13 +0000 (13:07 -0700)
egress_priority_map[] hash table updates are protected by rtnl,
and we never remove elements until device is dismantled.

We have to make sure that before inserting an new element in hash table,
all its fields are committed to memory or else another cpu could
find corrupt values and crash.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
net/8021q/vlan_dev.c

index 3a8c8fd63c88e2c204a8f1ac3b6dc4eb3f15cc63..1cd3d2a406f5a268113d0d802f03578f2a3cbe20 100644 (file)
@@ -73,6 +73,8 @@ vlan_dev_get_egress_qos_mask(struct net_device *dev, struct sk_buff *skb)
 {
        struct vlan_priority_tci_mapping *mp;
 
+       smp_rmb(); /* coupled with smp_wmb() in vlan_dev_set_egress_priority() */
+
        mp = vlan_dev_priv(dev)->egress_priority_map[(skb->priority & 0xF)];
        while (mp) {
                if (mp->priority == skb->priority) {
@@ -249,6 +251,11 @@ int vlan_dev_set_egress_priority(const struct net_device *dev,
        np->next = mp;
        np->priority = skb_prio;
        np->vlan_qos = vlan_qos;
+       /* Before inserting this element in hash table, make sure all its fields
+        * are committed to memory.
+        * coupled with smp_rmb() in vlan_dev_get_egress_qos_mask()
+        */
+       smp_wmb();
        vlan->egress_priority_map[skb_prio & 0xF] = np;
        if (vlan_qos)
                vlan->nr_egress_mappings++;