bpf: Set mac_len in bpf_skb_change_head

[ Upstream commit 84316ca4e100d8cbfccd9f774e23817cb2059868 ]

The skb_change_head() helper did not set "skb->mac_len", which is
problematic when it's used in combination with skb_redirect_peer().
Without it, redirecting a packet from a L3 device such as wireguard to
the veth peer device will cause skb->data to point to the middle of the
IP header on entry to tcp_v4_rcv() since the L2 header is not pulled
correctly due to mac_len=0.

Fixes: 3a0af8fd61f9 ("bpf: BPF for lightweight tunnel infrastructure")
Signed-off-by: Jussi Maki <joamaki@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210519154743.2554771-2-joamaki@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

diff --git a/net/core/filter.c b/net/core/filter.c
index 9358bc4a3711f0f98446a6488b1e5f1ef229b14e..ef6bdbb63ecbba0c140aa3a6138f73888a66a8eb 100644 (file)
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -3782,6 +3782,7 @@ static inline int __bpf_skb_change_head(struct sk_buff *skb, u32 head_room,
                __skb_push(skb, head_room);
                memset(skb->data, 0, head_room);
                skb_reset_mac_header(skb);
+               skb_reset_mac_len(skb);
        }
 
        return ret;