tcp: ensure skb->dev is NULL before leaving TCP stack

skb->rbnode is sharing three skb fields : next, prev, dev

When a packet is sent, TCP keeps the original skb (master)
in a rtx queue, which was converted to rbtree a while back.

__tcp_transmit_skb() is responsible to clone the master skb,
and add the TCP header to the clone before sending it
to network layer.

skb_clone() already clears skb->next and skb->prev, but copies
the master oskb->dev into the clone.

We need to clear skb->dev, otherwise lower layers could interpret
the value as a pointer to a netdev.

This old bug surfaced recently when commit 28f8bfd
("netfilter: Support iif matches in POSTROUTING") was merged.

Before this netfilter commit, skb->dev value was ignored and
changed before reaching dev_queue_xmit()

Fixes: 75c119a ("tcp: implement rb-tree based retransmit queue")
Fixes: 28f8bfd ("netfilter: Support iif matches in POSTROUTING")
Signed-off-by: Eric Dumazet <[email protected]>
Reported-by: Martin Zaharinov <[email protected]>
Cc: Florian Westphal <[email protected]>
Cc: Pablo Neira Ayuso <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Author: edumazet

net/ipv4/tcp_output.c

@@ -1109,6 +1109,10 @@ static int __tcp_transmit_skb(struct sock *sk, struct sk_buff *skb,
 
 		if (unlikely(!skb))
 			return -ENOBUFS;
+		/* retransmit skbs might have a non zero value in skb->dev
+		 * because skb->dev is aliased with skb->rbnode.rb_left
+		 */
+		skb->dev = NULL;
 	}
 
 	inet = inet_sk(sk);