[PATCH net] ipv4: Pass original device to ip_rcv_finish_core

[PATCH net] ipv4: Pass original device to ip_rcv_finish_core

[David Ahern]

From: David Ahern <dsahern@gmail.com>

ip_route_input_rcu expects the original ingress device (e.g., for
proper multicast handling). The skb->dev can be changed by l3mdev_ip_rcv,
so dev needs to be saved prior to calling it. This was the behavior prior
to the listify changes.

Fixes: 5fa12739a53d0 ("net: ipv4: listify ip_rcv_finish")
Cc: Edward Cree <ecree@solarflare.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
---
 net/ipv4/ip_input.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/net/ipv4/ip_input.c b/net/ipv4/ip_input.c
index 51d8efba6de2..1f4737b77067 100644
--- a/net/ipv4/ip_input.c
+++ b/net/ipv4/ip_input.c
@@ -307,11 +307,10 @@ static inline bool ip_rcv_options(struct sk_buff *skb)
 }
 
 static int ip_rcv_finish_core(struct net *net, struct sock *sk,
-			      struct sk_buff *skb)
+			      struct sk_buff *skb, struct net_device *dev)
 {
 	const struct iphdr *iph = ip_hdr(skb);
 	int (*edemux)(struct sk_buff *skb);
-	struct net_device *dev = skb->dev;
 	struct rtable *rt;
 	int err;
 
@@ -400,6 +399,7 @@ static int ip_rcv_finish_core(struct net *net, struct sock *sk,
 
 static int ip_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
 {
+	struct net_device *dev = skb->dev;
 	int ret;
 
 	/* if ingress device is enslaved to an L3 master device pass the
@@ -409,7 +409,7 @@ static int ip_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
 	if (!skb)
 		return NET_RX_SUCCESS;
 
-	ret = ip_rcv_finish_core(net, sk, skb);
+	ret = ip_rcv_finish_core(net, sk, skb, dev);
 	if (ret != NET_RX_DROP)
 		ret = dst_input(skb);
 	return ret;
@@ -545,6 +545,7 @@ static void ip_list_rcv_finish(struct net *net, struct sock *sk,
 
 	INIT_LIST_HEAD(&sublist);
 	list_for_each_entry_safe(skb, next, head, list) {
+		struct net_device *dev = skb->dev;
 		struct dst_entry *dst;
 
 		skb_list_del_init(skb);
@@ -554,7 +555,7 @@ static void ip_list_rcv_finish(struct net *net, struct sock *sk,
 		skb = l3mdev_ip_rcv(skb);
 		if (!skb)
 			continue;
-		if (ip_rcv_finish_core(net, sk, skb) == NET_RX_DROP)
+		if (ip_rcv_finish_core(net, sk, skb, dev) == NET_RX_DROP)
 			continue;
 
 		dst = skb_dst(skb);
-- 
2.11.0

Re: [PATCH net] ipv4: Pass original device to ip_rcv_finish_core

[David Miller]

From: David Ahern <dsahern@kernel.org>
Date: Mon, 25 Feb 2019 13:55:48 -0800

> From: David Ahern <dsahern@gmail.com>
> 
> ip_route_input_rcu expects the original ingress device (e.g., for
> proper multicast handling). The skb->dev can be changed by l3mdev_ip_rcv,
> so dev needs to be saved prior to calling it. This was the behavior prior
> to the listify changes.
> 
> Fixes: 5fa12739a53d0 ("net: ipv4: listify ip_rcv_finish")
> Cc: Edward Cree <ecree@solarflare.com>
> Signed-off-by: David Ahern <dsahern@gmail.com>

Applied and queued up for -stable, thanks Dave.