openvswitch: Use regular VXLAN net_device device

This gets rid of all OVS specific VXLAN code in the receive and
transmit path by using a VXLAN net_device to represent the vport.
Only a small shim layer remains which takes care of handling the
VXLAN specific OVS Netlink configuration.

Unexports vxlan_sock_add(), vxlan_sock_release(), vxlan_xmit_skb()
since they are no longer needed.

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 198 insertions, 3 deletions

diff --git a/net/openvswitch/vport-netdev.c b/net/openvswitch/vport-netdev.c
index e682bdc..68d0582 100644
--- a/net/openvswitch/vport-netdev.c
+++ b/net/openvswitch/vport-netdev.c
@@ -27,9 +27,13 @@
 #include <linux/skbuff.h>
 #include <linux/openvswitch.h>
 
-#include <net/llc.h>
+#include <net/udp.h>
+#include <net/ip_tunnels.h>
+#include <net/rtnetlink.h>
+#include <net/vxlan.h>
 
 #include "datapath.h"
+#include "vport.h"
 #include "vport-internal_dev.h"
 #include "vport-netdev.h"
 
@@ -147,7 +151,8 @@ static void free_port_rcu(struct rcu_head *rcu)
 {
 	struct vport *vport = container_of(rcu, struct vport, rcu);
 
-	dev_put(vport->dev);
+	if (vport->dev)
+		dev_put(vport->dev);
 	ovs_vport_free(vport);
 }
 
@@ -221,12 +226,202 @@ static struct vport_ops ovs_netdev_vport_ops = {
 	.send		= netdev_send,
 };
 
+/* Compat code for old userspace. */
+#if IS_ENABLED(CONFIG_VXLAN)
+static struct vport_ops ovs_vxlan_netdev_vport_ops;
+
+static int vxlan_get_options(const struct vport *vport, struct sk_buff *skb)
+{
+	struct vxlan_dev *vxlan = netdev_priv(vport->dev);
+	__be16 dst_port = vxlan->cfg.dst_port;
+
+	if (nla_put_u16(skb, OVS_TUNNEL_ATTR_DST_PORT, ntohs(dst_port)))
+		return -EMSGSIZE;
+
+	if (vxlan->flags & VXLAN_F_GBP) {
+		struct nlattr *exts;
+
+		exts = nla_nest_start(skb, OVS_TUNNEL_ATTR_EXTENSION);
+		if (!exts)
+			return -EMSGSIZE;
+
+		if (vxlan->flags & VXLAN_F_GBP &&
+		    nla_put_flag(skb, OVS_VXLAN_EXT_GBP))
+			return -EMSGSIZE;
+
+		nla_nest_end(skb, exts);
+	}
+
+	return 0;
+}
+
+static const struct nla_policy exts_policy[OVS_VXLAN_EXT_MAX + 1] = {
+	[OVS_VXLAN_EXT_GBP]	= { .type = NLA_FLAG, },
+};
+
+static int vxlan_configure_exts(struct vport *vport, struct nlattr *attr,
+				struct vxlan_config *conf)
+{
+	struct nlattr *exts[OVS_VXLAN_EXT_MAX + 1];
+	int err;
+
+	if (nla_len(attr) < sizeof(struct nlattr))
+		return -EINVAL;
+
+	err = nla_parse_nested(exts, OVS_VXLAN_EXT_MAX, attr, exts_policy);
+	if (err < 0)
+		return err;
+
+	if (exts[OVS_VXLAN_EXT_GBP])
+		conf->flags |= VXLAN_F_GBP;
+
+	return 0;
+}
+
+static struct vport *vxlan_tnl_create(const struct vport_parms *parms)
+{
+	struct net *net = ovs_dp_get_net(parms->dp);
+	struct nlattr *options = parms->options;
+	struct net_device *dev;
+	struct vport *vport;
+	struct nlattr *a;
+	int err;
+	struct vxlan_config conf = {
+		.no_share = true,
+		.flags = VXLAN_F_FLOW_BASED | VXLAN_F_COLLECT_METADATA,
+	};
+
+	if (!options) {
+		err = -EINVAL;
+		goto error;
+	}
+
+	a = nla_find_nested(options, OVS_TUNNEL_ATTR_DST_PORT);
+	if (a && nla_len(a) == sizeof(u16)) {
+		conf.dst_port = htons(nla_get_u16(a));
+	} else {
+		/* Require destination port from userspace. */
+		err = -EINVAL;
+		goto error;
+	}
+
+	vport = ovs_vport_alloc(0, &ovs_vxlan_netdev_vport_ops, parms);
+	if (IS_ERR(vport))
+		return vport;
+
+	a = nla_find_nested(options, OVS_TUNNEL_ATTR_EXTENSION);
+	if (a) {
+		err = vxlan_configure_exts(vport, a, &conf);
+		if (err) {
+			ovs_vport_free(vport);
+			goto error;
+		}
+	}
+
+	rtnl_lock();
+	dev = vxlan_dev_create(net, parms->name, NET_NAME_USER, &conf);
+	if (IS_ERR(dev)) {
+		rtnl_unlock();
+		ovs_vport_free(vport);
+		return ERR_CAST(dev);
+	}
+
+	dev_change_flags(dev, dev->flags | IFF_UP);
+	rtnl_unlock();
+	return vport;
+error:
+	return ERR_PTR(err);
+}
+
+static struct vport *vxlan_create(const struct vport_parms *parms)
+{
+	struct vport *vport;
+
+	vport = vxlan_tnl_create(parms);
+	if (IS_ERR(vport))
+		return vport;
+
+	return netdev_link(vport, parms->name);
+}
+
+static void vxlan_destroy(struct vport *vport)
+{
+	rtnl_lock();
+	if (vport->dev->priv_flags & IFF_OVS_DATAPATH)
+		ovs_netdev_detach_dev(vport);
+
+	/* Early release so we can unregister the device */
+	dev_put(vport->dev);
+	rtnl_delete_link(vport->dev);
+	vport->dev = NULL;
+	rtnl_unlock();
+
+	call_rcu(&vport->rcu, free_port_rcu);
+}
+
+static int vxlan_get_egress_tun_info(struct vport *vport, struct sk_buff *skb,
+				     struct ip_tunnel_info *egress_tun_info)
+{
+	struct vxlan_dev *vxlan = netdev_priv(vport->dev);
+	struct net *net = ovs_dp_get_net(vport->dp);
+	__be16 dst_port = vxlan_dev_dst_port(vxlan);
+	__be16 src_port;
+	int port_min;
+	int port_max;
+
+	inet_get_local_port_range(net, &port_min, &port_max);
+	src_port = udp_flow_src_port(net, skb, 0, 0, true);
+
+	return ovs_tunnel_get_egress_info(egress_tun_info, net,
+					  OVS_CB(skb)->egress_tun_info,
+					  IPPROTO_UDP, skb->mark,
+					  src_port, dst_port);
+}
+
+static struct vport_ops ovs_vxlan_netdev_vport_ops = {
+	.type		= OVS_VPORT_TYPE_VXLAN,
+	.create		= vxlan_create,
+	.destroy	= vxlan_destroy,
+	.get_options	= vxlan_get_options,
+	.send		= netdev_send,
+	.get_egress_tun_info	= vxlan_get_egress_tun_info,
+};
+
+static int vxlan_compat_init(void)
+{
+	return ovs_vport_ops_register(&ovs_vxlan_netdev_vport_ops);
+}
+
+static void vxlan_compat_exit(void)
+{
+	ovs_vport_ops_unregister(&ovs_vxlan_netdev_vport_ops);
+}
+#else
+static int vxlan_compat_init(void)
+{
+	return 0;
+}
+
+static void vxlan_compat_exit(void)
+{
+}
+#endif
+
 int __init ovs_netdev_init(void)
 {
-	return ovs_vport_ops_register(&ovs_netdev_vport_ops);
+	int err;
+
+	err = ovs_vport_ops_register(&ovs_netdev_vport_ops);
+	if (err)
+		return err;
+	err = vxlan_compat_init();
+	if (err)
+		vxlan_compat_exit();
+	return err;
 }
 
 void ovs_netdev_exit(void)
 {
 	ovs_vport_ops_unregister(&ovs_netdev_vport_ops);
+	vxlan_compat_exit();
 }