dev_forward_skb: do not scrub skb mark within the same name space

The goal is to keep the mark during a bpf_redirect(), like it is done for
legacy encapsulation / decapsulation, when there is no x-netns.
This was initially done in commit 213dd74 ("skbuff: Do not scrub skb
mark within the same name space").

When the call to skb_scrub_packet() was added in dev_forward_skb() (commit
8b27f27 ("skb: allow skb_scrub_packet() to be used by tunnels")), the
second argument (xnet) was set to true to force a call to skb_orphan(). At
this time, the mark was always cleanned up by skb_scrub_packet(), whatever
xnet value was.
This call to skb_orphan() was removed later in commit
9c4c325 ("skbuff: preserve sock reference when scrubbing the skb.").
But this 'true' stayed here without any real reason.

Let's correctly set xnet in ____dev_forward_skb(), this function has access
to the previous interface and to the new interface.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/netdevice.h

@@ -4114,7 +4114,7 @@ static __always_inline int ____dev_forward_skb(struct net_device *dev,
 		return NET_RX_DROP;
 	}
 
-	skb_scrub_packet(skb, true);
+	skb_scrub_packet(skb, !net_eq(dev_net(dev), dev_net(skb->dev)));
 	skb->priority = 0;
 	return 0;
 }