[PATCH v1 net 5/5] ip6_tunnel: Pull header after checking skb->protocol in ip6_tnl_start_xmit().

[Kuniyuki Iwashima <kuniyu@amazon.com>]

syzkaller demonstrated the underflow in pskb_network_may_pull()
by sending a crafted VLAN packet over tunnel devices: sit, ipip,
vti, and vti6.

The same warning will be triggered for ip6tnl, so let's check
skb->protocol before pulling the next header in ip6_tnl_start_xmit().

Fixes: cb9f1b783850 ("ip: validate header length on virtual device xmit")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
---
 net/ipv6/ip6_tunnel.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index e9cc315832cb..81be7a5be6c5 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -1424,14 +1424,17 @@ ip6_tnl_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	u8 ipproto;
 	int ret;
 
-	if (!pskb_inet_may_pull(skb))
-		goto tx_err;
-
 	switch (skb->protocol) {
 	case htons(ETH_P_IP):
+		if (!pskb_network_may_pull(skb, sizeof(struct iphdr)))
+			goto tx_err;
+
 		ipproto = IPPROTO_IPIP;
 		break;
 	case htons(ETH_P_IPV6):
+		if (!pskb_network_may_pull(skb, sizeof(struct ipv6hdr)))
+			goto tx_err;
+
 		if (ip6_tnl_addr_conflict(t, ipv6_hdr(skb)))
 			goto tx_err;
 		ipproto = IPPROTO_IPV6;
-- 
2.30.2