[PATCH 5.10,5.15] ip6_tunnel: Fix usage of skb_vlan_inet_prepare()

[PATCH 5.10,5.15] ip6_tunnel: Fix usage of skb_vlan_inet_prepare()

[Ben Hutchings]

[-- Attachment #1: Type: text/plain, Size: 1338 bytes --]

Backports of commit 81c734dae203 "ip6_tunnel: use
skb_vlan_inet_prepare() in __ip6_tnl_rcv()" broke IPv6 tunnelling in
stable branches 5.10-6.12 inclusive.  This is because the return value
of skb_vlan_inet_prepare() had the opposite sense (0 for error rather
than for success) before commit 9990ddf47d416 "net: tunnel: make
skb_vlan_inet_prepare() return drop reasons".

For branches including commit c504e5c2f964 "net: skb: introduce
kfree_skb_reason()" etc. (i.e. 6.1 and newer) it was simple to
backport commit 9990ddf47d416, but for 5.10 and 5.15 that doesn't seem
to be practical.

So just reverse the sense of the return value test here.

Fixes: f9c5c5b791d3 ("ip6_tunnel: use skb_vlan_inet_prepare() in __ip6_tnl_rcv()")
Fixes: 64c71d60a21a ("ip6_tunnel: use skb_vlan_inet_prepare() in __ip6_tnl_rcv()")
Signed-off-by: Ben Hutchings <benh@debian.org>
---
I built and tested 5.10.251 with this on top, but have not tested it
with the 5.15 branch.

Ben.

--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -876,7 +876,7 @@ static int __ip6_tnl_rcv(struct ip6_tnl
 
 	skb_reset_network_header(skb);
 
-	if (skb_vlan_inet_prepare(skb, true)) {
+	if (!skb_vlan_inet_prepare(skb, true)) {
 		DEV_STATS_INC(tunnel->dev, rx_length_errors);
 		DEV_STATS_INC(tunnel->dev, rx_errors);
 		goto drop;

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]