If OVS receives a packet from another namespace, then the packet should
be scrubbed. However, people have already begun to rely on the behaviour
that skb->mark is preserved across namespaces, so retain this one field.
This is mainly to address information leakage between namespaces when
using OVS internal ports, but by placing it in ovs_vport_receive() it is
more generally applicable, meaning it should not be overlooked if other
port types are allowed to be moved into namespaces in future.
Signed-off-by: Joe Stringer <redacted>
---
I originally proposed this patch as part of the conntrack changes to OVS,
and there was some discussion on that thread, culminating here:
http://www.spinics.net/lists/netdev/msg338626.html
We also discussed this a bit in Seattle, however I didn't follow up
immediately so I don't exactly recall what the consensus was. Following
Jesse's direction in the above thread, I'm proposing that we preserve the
mark, but scrub the rest. Also fixed the use-after-free bug present in the
previous version.
I think this is relevant for 'net', because this is the first time that
the metadata_dst and nfct are exposed (albeit indirectly) through OVS so it
would be nice to get agreement on the expected behaviour.
---
net/openvswitch/vport.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/net/openvswitch/vport.c b/net/openvswitch/vport.c
index fc5c0b9ccfe9..70f19ea99b92 100644
--- a/net/openvswitch/vport.c
+++ b/net/openvswitch/vport.c
@@ -440,10 +440,17 @@ int ovs_vport_receive(struct vport *vport, struct sk_buff *skb,
const struct ip_tunnel_info *tun_info)
{
struct sw_flow_key key;
+ u32 mark = skb->mark;
int error;
OVS_CB(skb)->input_vport = vport;
OVS_CB(skb)->mru = 0;
+ if (dev_net(skb->dev) != ovs_dp_get_net(vport->dp)) {
+ skb_scrub_packet(skb, true);
+ tun_info = NULL;
+ }
+ skb->mark = mark;
+
/* Extract flow from 'skb' into 'key'. */
error = ovs_flow_key_extract(tun_info, skb, &key);
if (unlikely(error)) {--
2.1.4