Re: [PATCH net-next 5/9] openvswitch: Add conntrack action

Thanks for review,

On 31 July 2015 at 07:52, Hannes Frederic Sowa [off-list ref] wrote:

On Thu, 2015-07-30 at 11:12 -0700, Joe Stringer wrote:

quoted

+static void prepare_frag(struct vport *vport, struct sw_flow_key
*key,
+                      struct sk_buff *skb)
+{
+     unsigned int hlen = ETH_HLEN;
+     struct ovs_frag_data *data;
+
+     data = this_cpu_ptr(&ovs_frag_data_storage);
+     data->dst = skb_dst(skb);

If data->dst is unsigned long, we could simply use an assignment:

data->dst = skb->_skb_refdst;

At this point we never leave rcu_read_lock section, so we are safe,
maybe we can add a comment for that.

OK, it also may be helpful to highlight that prepare_frag() is done
once for an assembled frame, then ovs_vport_output() performs the
inverse for each fragment. I'll do this.

...

quoted

+             } else if (key->eth.type == htons(ETH_P_IP)) {
+                     struct dst_entry ovs_dst;
+
+                     prepare_frag(vport, key, skb);
+                     dst_init(&ovs_dst, &ovs_dst_ops, vport->dev,
+                              1, DST_OBSOLETE_NONE, DST_NOCOUNT);

I don't think we should take a ref on the netdev here.

dst_init(&ovs_dst, &ovs_dst_ops, NULL,
         1, DST_OBSOLETE_NONE, DST_NOCOUNT);
ovs_dst.dev = vport->dev;

Some of this was me being overly cautious: take a ref on the dev for
as long as the fragment dst exists; take a ref on the original (eg
tunnel_metadata) dst for the length of handling the output of this
frame.

quoted

+
+                     skb_dst_drop(skb);
+                     skb_dst_set_noref(skb, &ovs_dst);
+                     IPCB(skb)->frag_max_size = mru;
+
+                     ip_do_fragment(skb->sk, skb,
ovs_vport_output);
+                     dev_put(ovs_dst.dev);

Can be removed then.

It seems a little strange to leave the skb->dst attached to the skb but
drop the reference from the netdevice here. Maybe a comment would make
sense, otherwise it smells fishy.

For each fragment, ovs_vport_output() will revert the changes made
here - restoring the original dst. Either way, if we're not taking a
ref on the netdev then this should be fine.