net: shy netns_ok check
Alexey Dobriyan [Thu, 13 Nov 2008 07:23:51 +0000 (23:23 -0800)]
Failure to pass netns_ok check is SILENT, except some MIB counter is
incremented somewhere.

And adding "netns_ok = 1" (after long head-scratching session) is
usually the last step in making some protocol netns-ready...

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/ipv4/ip_input.c

index 861978a..cfb38ac 100644 (file)
@@ -209,9 +209,17 @@ static int ip_local_deliver_finish(struct sk_buff *skb)
 
                hash = protocol & (MAX_INET_PROTOS - 1);
                ipprot = rcu_dereference(inet_protos[hash]);
-               if (ipprot != NULL && (net == &init_net || ipprot->netns_ok)) {
+               if (ipprot != NULL) {
                        int ret;
 
+                       if (!net_eq(net, &init_net) && !ipprot->netns_ok) {
+                               if (net_ratelimit())
+                                       printk("%s: proto %d isn't netns-ready\n",
+                                               __func__, protocol);
+                               kfree_skb(skb);
+                               goto out;
+                       }
+
                        if (!ipprot->no_policy) {
                                if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) {
                                        kfree_skb(skb);