datapath: Set a large MTU on tunnel devices.

Upstream commit:
    Prior to 4.3, openvswitch tunnel vports (vxlan, gre and geneve) could
    transmit vxlan packets of any size, constrained only by the ability to
    send out the resulting packets.  4.3 introduced netdevs corresponding
    to tunnel vports.  These netdevs have an MTU, which limits the size of
    a packet that can be successfully encapsulated.  The default MTU
    values are low (1500 or less), which is awkwardly small in the context
    of physical networks supporting jumbo frames, and leads to a
    conspicuous change in behaviour for userspace.

    Instead, set the MTU on openvswitch-created netdevs to be the relevant
    maximum (i.e. the maximum IP packet size minus any relevant overhead),
    effectively restoring the behaviour prior to 4.3.

    Signed-off-by: David Wragg <david@weave.works>
    Signed-off-by: David S. Miller <davem@davemloft.net>

Upstream: 7e059158d57b ("vxlan, gre, geneve: Set a large MTU on ovs-created
tunnel devices")
Signed-off-by: Joe Stringer <joe@ovn.org>
Acked-by: Jesse Gross <jesse@kernel.org>

acinclude.m4

@@ -354,6 +354,7 @@ AC_DEFUN([OVS_CHECK_LINUX_COMPAT], [
 
   OVS_GREP_IFELSE([$KSRC/include/net/ip.h], [IPSKB_FRAG_PMTU],
                   [OVS_DEFINE([HAVE_CORRECT_MRU_HANDLING])])
+  OVS_GREP_IFELSE([$KSRC/include/net/ip_tunnels.h], [__ip_tunnel_change_mtu])
   OVS_GREP_IFELSE([$KSRC/include/net/inet_frag.h], [hashfn.*const],
                   [OVS_DEFINE([HAVE_INET_FRAGS_CONST])])
   OVS_GREP_IFELSE([$KSRC/include/net/inet_frag.h], [last_in],

datapath/linux/compat/geneve.c

@@ -1039,11 +1039,21 @@ struct net_device *rpl_geneve_dev_create_fb(struct net *net, const char *name,
 		return dev;
 
 	err = geneve_configure(net, dev, 0, 0, 0, 0, htons(dst_port), true);
-	if (err) {
-		free_netdev(dev);
-		return ERR_PTR(err);
-	}
+	if (err)
+		goto err;
+
+	/* openvswitch users expect packet sizes to be unrestricted,
+	 * so set the largest MTU we can.
+	 */
+	err = geneve_change_mtu(dev, IP_MAX_MTU);
+	if (err)
+		goto err;
+
 	return dev;
+
+err:
+	free_netdev(dev);
+	return ERR_PTR(err);
 }
 EXPORT_SYMBOL_GPL(rpl_geneve_dev_create_fb);
 

datapath/linux/compat/include/net/ip_tunnels.h

@@ -327,4 +327,10 @@ int rpl_ip_tunnel_get_iflink(const struct net_device *dev);
 #define ip_tunnel_get_link_net rpl_ip_tunnel_get_link_net
 struct net *rpl_ip_tunnel_get_link_net(const struct net_device *dev);
 #endif /* HAVE_METADATA_DST */
+
+#ifndef HAVE___IP_TUNNEL_CHANGE_MTU
+#define __ip_tunnel_change_mtu rpl___ip_tunnel_change_mtu
+int rpl___ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict);
+#endif
+
 #endif /* __NET_IP_TUNNELS_H */

datapath/linux/compat/ip_gre.c

@@ -613,6 +613,14 @@ struct net_device *rpl_gretap_fb_dev_create(struct net *net, const char *name,
 #endif
 	if (err < 0)
 		goto out;
+
+	/* openvswitch users expect packet sizes to be unrestricted,
+	 * so set the largest MTU we can.
+	 */
+	err = __ip_tunnel_change_mtu(dev, IP_MAX_MTU, false);
+	if (err)
+		goto out;
+
 	return dev;
 out:
 	free_netdev(dev);

datapath/linux/compat/ip_tunnel.c

@@ -137,18 +137,31 @@ static int ip_tunnel_bind_dev(struct net_device *dev)
 	return mtu;
 }
 
-int rpl_ip_tunnel_change_mtu(struct net_device *dev, int new_mtu)
+int rpl___ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict)
 {
 	struct ip_tunnel *tunnel = netdev_priv(dev);
 	int t_hlen = tunnel->hlen + sizeof(struct iphdr);
+	int max_mtu = 0xFFF8 - dev->hard_header_len - t_hlen;
 
-	if (new_mtu < 68 ||
-	    new_mtu > 0xFFF8 - dev->hard_header_len - t_hlen)
+	if (new_mtu < 68)
 		return -EINVAL;
+
+	if (new_mtu > max_mtu) {
+		if (strict)
+			return -EINVAL;
+
+		new_mtu = max_mtu;
+	}
+
 	dev->mtu = new_mtu;
 	return 0;
 }
 
+int rpl_ip_tunnel_change_mtu(struct net_device *dev, int new_mtu)
+{
+	return rpl___ip_tunnel_change_mtu(dev, new_mtu, true);
+}
+
 static void ip_tunnel_dev_free(struct net_device *dev)
 {
 #ifdef HAVE_DEV_TSTATS

datapath/linux/compat/vxlan.c

@@ -1905,6 +1905,7 @@ static int vxlan_dev_configure(struct net *src_net, struct net_device *dev,
 	int err;
 	bool use_ipv6 = false;
 	__be16 default_port = vxlan->cfg.dst_port;
+	struct net_device *lowerdev = NULL;
 
 	vxlan->net = src_net;
 
@@ -1924,9 +1925,7 @@ static int vxlan_dev_configure(struct net *src_net, struct net_device *dev,
 	}
 
 	if (conf->remote_ifindex) {
-		struct net_device *lowerdev
-			 = __dev_get_by_index(src_net, conf->remote_ifindex);
-
+		lowerdev = __dev_get_by_index(src_net, conf->remote_ifindex);
 		dst->remote_ifindex = conf->remote_ifindex;
 
 		if (!lowerdev) {
@@ -1957,6 +1956,12 @@ static int vxlan_dev_configure(struct net *src_net, struct net_device *dev,
 		dev->needed_headroom = ETH_HLEN + VXLAN_HEADROOM;
 	}
 
+	if (conf->mtu) {
+		err = __vxlan_change_mtu(dev, lowerdev, dst, conf->mtu, false);
+		if (err)
+			return err;
+	}
+
 	memcpy(&vxlan->cfg, conf, sizeof(*conf));
 	if (!vxlan->cfg.dst_port)
 		vxlan->cfg.dst_port = default_port;

datapath/vport-vxlan.c

@@ -91,6 +91,8 @@ static struct vport *vxlan_tnl_create(const struct vport_parms *parms)
 	struct vxlan_config conf = {
 		.no_share = true,
 		.flags = VXLAN_F_COLLECT_METADATA,
+		/* Don't restrict the packets that can be sent by MTU */
+		.mtu = IP_MAX_MTU,
 	};
 
 	if (!options) {