vti: do not confirm neighbor when do pmtu update
authorHangbin Liu <liuhangbin@gmail.com>
Sun, 22 Dec 2019 02:51:14 +0000 (10:51 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 4 Jan 2020 13:00:17 +0000 (14:00 +0100)
[ Upstream commit 8247a79efa2f28b44329f363272550c1738377de ]

When do IPv6 tunnel PMTU update and calls __ip6_rt_update_pmtu() in the end,
we should not call dst_confirm_neigh() as there is no two-way communication.

Although vti and vti6 are immune to this problem because they are IFF_NOARP
interfaces, as Guillaume pointed. There is still no sense to confirm neighbour
here.

v5: Update commit description.
v4: No change.
v3: Do not remove dst_confirm_neigh, but add a new bool parameter in
    dst_ops.update_pmtu to control whether we should do neighbor confirm.
    Also split the big patch to small ones for each area.
v2: Remove dst_confirm_neigh in __ip6_rt_update_pmtu.

Reviewed-by: Guillaume Nault <gnault@redhat.com>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/ipv4/ip_vti.c
net/ipv6/ip6_vti.c

index c07065b7e3b0e1c045812a153ce721c751ac8354..08c15dd42d935fd95190d5469bf7e805769fb8e3 100644 (file)
@@ -244,7 +244,7 @@ static netdev_tx_t vti_xmit(struct sk_buff *skb, struct net_device *dev,
 
        mtu = dst_mtu(dst);
        if (skb->len > mtu) {
-               skb_dst_update_pmtu(skb, mtu);
+               skb_dst_update_pmtu_no_confirm(skb, mtu);
                if (skb->protocol == htons(ETH_P_IP)) {
                        icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED,
                                  htonl(mtu));
index 6b2416b4a53e53ac574a126fbd9bc5c978188f65..557fe3880a3f37082309ffd7fddba61ed95a6919 100644 (file)
@@ -483,7 +483,7 @@ vti6_xmit(struct sk_buff *skb, struct net_device *dev, struct flowi *fl)
 
        mtu = dst_mtu(dst);
        if (skb->len > mtu) {
-               skb_dst_update_pmtu(skb, mtu);
+               skb_dst_update_pmtu_no_confirm(skb, mtu);
 
                if (skb->protocol == htons(ETH_P_IPV6)) {
                        if (mtu < IPV6_MIN_MTU)