[PATCH net-next v3 6/6] openvswitch: Add vxlan tunneling support.

[PATCH net-next v3 6/6] openvswitch: Add vxlan tunneling support.

[Pravin B Shelar]

Following patch adds vxlan vport type for openvswitch using
vxlan api.
This patch adds vxlan dependency for openvswitch.

CC: Jesse Gross <jesse@nicira.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
---
v2-v3:
 - Moved CONFIG_OPENVSWITCH_VXLAN check to makefile.
v1-v2:
 - removed NULL check in vxlan-rcv.
 - change vxlan-port create to look good.
 - Added Cisco copyright.
---
 include/uapi/linux/openvswitch.h |   11 ++
 net/openvswitch/Kconfig          |   13 +++
 net/openvswitch/Makefile         |    4 +
 net/openvswitch/vport-vxlan.c    |  208 ++++++++++++++++++++++++++++++++++++++
 net/openvswitch/vport.c          |    3 +
 net/openvswitch/vport.h          |    1 +
 6 files changed, 240 insertions(+), 0 deletions(-)
 create mode 100644 net/openvswitch/vport-vxlan.c

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h
index c55efaa..52490b0 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -165,6 +165,7 @@ enum ovs_vport_type {
 	OVS_VPORT_TYPE_NETDEV,   /* network device */
 	OVS_VPORT_TYPE_INTERNAL, /* network device implemented by datapath */
 	OVS_VPORT_TYPE_GRE,      /* GRE tunnel. */
+	OVS_VPORT_TYPE_VXLAN,	 /* VXLAN tunnel. */
 	__OVS_VPORT_TYPE_MAX
 };
 
@@ -211,6 +212,16 @@ enum ovs_vport_attr {
 
 #define OVS_VPORT_ATTR_MAX (__OVS_VPORT_ATTR_MAX - 1)
 
+/* OVS_VPORT_ATTR_OPTIONS attributes for tunnels.
+ */
+enum {
+	OVS_TUNNEL_ATTR_UNSPEC,
+	OVS_TUNNEL_ATTR_DST_PORT, /* 16-bit UDP port, used by L4 tunnels. */
+	__OVS_TUNNEL_ATTR_MAX
+};
+
+#define OVS_TUNNEL_ATTR_MAX (__OVS_TUNNEL_ATTR_MAX - 1)
+
 /* Flows. */
 
 #define OVS_FLOW_FAMILY  "ovs_flow"
diff --git a/net/openvswitch/Kconfig b/net/openvswitch/Kconfig
index 27ee56b..bed30e6 100644
--- a/net/openvswitch/Kconfig
+++ b/net/openvswitch/Kconfig
@@ -40,3 +40,16 @@ config OPENVSWITCH_GRE
 	  Say N to exclude this support and reduce the binary size.
 
 	  If unsure, say Y.
+
+config OPENVSWITCH_VXLAN
+	bool "Open vSwitch VXLAN tunneling support"
+	depends on INET
+	depends on OPENVSWITCH
+	depends on VXLAN && !(OPENVSWITCH=y && VXLAN=m)
+	default y
+	---help---
+	  If you say Y here, then the Open vSwitch will be able create vxlan vport.
+
+	  Say N to exclude this support and reduce the binary size.
+
+	  If unsure, say Y.
diff --git a/net/openvswitch/Makefile b/net/openvswitch/Makefile
index 01bddb2..82e4ee5 100644
--- a/net/openvswitch/Makefile
+++ b/net/openvswitch/Makefile
@@ -13,3 +13,7 @@ openvswitch-y := \
 	vport-gre.o \
 	vport-internal_dev.o \
 	vport-netdev.o
+
+ifneq ($(CONFIG_OPENVSWITCH_VXLAN),)
+openvswitch-y += vport-vxlan.o
+endif
diff --git a/net/openvswitch/vport-vxlan.c b/net/openvswitch/vport-vxlan.c
new file mode 100644
index 0000000..ad9af2d
--- /dev/null
+++ b/net/openvswitch/vport-vxlan.c
@@ -0,0 +1,208 @@
+/*
+ * Copyright (c) 2013 Nicira, Inc.
+ * Copyright (c) 2013 Cisco Systems, Inc.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of version 2 of the GNU General Public
+ * License as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/in.h>
+#include <linux/ip.h>
+#include <linux/net.h>
+#include <linux/rculist.h>
+#include <linux/udp.h>
+
+#include <net/icmp.h>
+#include <net/ip.h>
+#include <net/udp.h>
+#include <net/ip_tunnels.h>
+#include <net/udp.h>
+#include <net/rtnetlink.h>
+#include <net/route.h>
+#include <net/dsfield.h>
+#include <net/inet_ecn.h>
+#include <net/net_namespace.h>
+#include <net/netns/generic.h>
+#include <net/vxlan.h>
+
+#include "datapath.h"
+#include "vport.h"
+
+#define OVS_VXLAN_RCV_PRIORITY		8
+
+/**
+ * struct vxlan_port - Keeps track of open UDP ports
+ * @vh: vxlan_handler created for the port.
+ * @name: vport name.
+ */
+struct vxlan_port {
+	struct vxlan_handler *vh;
+	char name[IFNAMSIZ];
+};
+
+static inline struct vxlan_port *vxlan_vport(const struct vport *vport)
+{
+	return vport_priv(vport);
+}
+
+/* Called with rcu_read_lock and BH disabled. */
+static int vxlan_rcv(struct vxlan_handler *vh, struct sk_buff *skb, __be32 vx_vni)
+{
+	struct ovs_key_ipv4_tunnel tun_key;
+	struct vport *vport = vh->data;
+	struct iphdr *iph;
+	__be64 key;
+
+	/* Save outer tunnel values */
+	iph = ip_hdr(skb);
+	key = cpu_to_be64(ntohl(vx_vni) >> 8);
+	ovs_flow_tun_key_init(&tun_key, iph, key, TUNNEL_KEY);
+
+	ovs_vport_receive(vport, skb, &tun_key);
+	return PACKET_RCVD;
+}
+
+static int vxlan_get_options(const struct vport *vport, struct sk_buff *skb)
+{
+	struct vxlan_port *vxlan_port = vxlan_vport(vport);
+	__be16 dst_port = inet_sk(vxlan_port->vh->vs->sock->sk)->inet_sport;
+
+	if (nla_put_u16(skb, OVS_TUNNEL_ATTR_DST_PORT, ntohs(dst_port)))
+		return -EMSGSIZE;
+	return 0;
+}
+
+static void vxlan_tnl_destroy(struct vport *vport)
+{
+	struct vxlan_port *vxlan_port = vxlan_vport(vport);
+
+	vxlan_handler_put(vxlan_port->vh);
+
+	ovs_vport_deferred_free(vport);
+}
+
+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 vxlan_port *vxlan_port;
+	struct vxlan_handler *vh;
+	struct vport *vport;
+	struct nlattr *a;
+	u16 dst_port;
+	int err;
+
+	if (!options) {
+		err = -EINVAL;
+		goto error;
+	}
+	a = nla_find_nested(options, OVS_TUNNEL_ATTR_DST_PORT);
+	if (a && nla_len(a) == sizeof(u16)) {
+		dst_port = nla_get_u16(a);
+	} else {
+		/* Require destination port from userspace. */
+		err = -EINVAL;
+		goto error;
+	}
+
+	vport = ovs_vport_alloc(sizeof(struct vxlan_port),
+				&ovs_vxlan_vport_ops, parms);
+	if (IS_ERR(vport))
+		return vport;
+
+	vxlan_port = vxlan_vport(vport);
+	strncpy(vxlan_port->name, parms->name, IFNAMSIZ);
+
+	vh = vxlan_handler_add(net, htons(dst_port), vxlan_rcv,
+			       vport, OVS_VXLAN_RCV_PRIORITY, true);
+	if (IS_ERR(vh)) {
+		ovs_vport_free(vport);
+		return (void *)vh;
+	}
+	vxlan_port->vh = vh;
+
+	return vport;
+
+error:
+	return ERR_PTR(err);
+}
+
+static int vxlan_tnl_send(struct vport *vport, struct sk_buff *skb)
+{
+	struct net *net = ovs_dp_get_net(vport->dp);
+	struct vxlan_port *vxlan_port = vxlan_vport(vport);
+	__be16 dst_port = inet_sk(vxlan_port->vh->vs->sock->sk)->inet_sport;
+	struct rtable *rt;
+	struct flowi4 fl;
+	__be16 src_port;
+	int port_min;
+	int port_max;
+	__be16 df;
+	int err;
+
+	if (unlikely(!OVS_CB(skb)->tun_key)) {
+		err = -EINVAL;
+		goto error;
+	}
+
+	/* Route lookup */
+	memset(&fl, 0, sizeof(fl));
+	fl.daddr = OVS_CB(skb)->tun_key->ipv4_dst;
+	fl.saddr = OVS_CB(skb)->tun_key->ipv4_src;
+	fl.flowi4_tos = RT_TOS(OVS_CB(skb)->tun_key->ipv4_tos);
+	fl.flowi4_mark = skb->mark;
+	fl.flowi4_proto = IPPROTO_UDP;
+
+	rt = ip_route_output_key(net, &fl);
+	if (IS_ERR(rt)) {
+		err = PTR_ERR(rt);
+		goto error;
+	}
+
+	df = OVS_CB(skb)->tun_key->tun_flags & TUNNEL_DONT_FRAGMENT ?
+		htons(IP_DF) : 0;
+
+	skb->local_df = 1;
+
+	inet_get_local_port_range(&port_min, &port_max);
+	src_port = vxlan_src_port(port_min, port_max, skb);
+
+	err = vxlan_xmit_skb(net, vxlan_port->vh, rt, skb,
+			     fl.saddr, OVS_CB(skb)->tun_key->ipv4_dst,
+			     OVS_CB(skb)->tun_key->ipv4_tos,
+			     OVS_CB(skb)->tun_key->ipv4_ttl, df,
+			     src_port, dst_port,
+			     htonl(be64_to_cpu(OVS_CB(skb)->tun_key->tun_id) << 8));
+	if (err < 0)
+		ip_rt_put(rt);
+error:
+	return err;
+}
+
+static const char *vxlan_get_name(const struct vport *vport)
+{
+	struct vxlan_port *vxlan_port = vxlan_vport(vport);
+	return vxlan_port->name;
+}
+
+const struct vport_ops ovs_vxlan_vport_ops = {
+	.type		= OVS_VPORT_TYPE_VXLAN,
+	.create		= vxlan_tnl_create,
+	.destroy	= vxlan_tnl_destroy,
+	.get_name	= vxlan_get_name,
+	.get_options	= vxlan_get_options,
+	.send		= vxlan_tnl_send,
+};
diff --git a/net/openvswitch/vport.c b/net/openvswitch/vport.c
index d4c7fa0..d69e0c0 100644
--- a/net/openvswitch/vport.c
+++ b/net/openvswitch/vport.c
@@ -42,6 +42,9 @@ static const struct vport_ops *vport_ops_list[] = {
 #ifdef CONFIG_OPENVSWITCH_GRE
 	&ovs_gre_vport_ops,
 #endif
+#ifdef CONFIG_OPENVSWITCH_VXLAN
+	&ovs_vxlan_vport_ops,
+#endif
 };
 
 /* Protected by RCU read lock for reading, ovs_mutex for writing. */
diff --git a/net/openvswitch/vport.h b/net/openvswitch/vport.h
index 376045c..1a9fbce 100644
--- a/net/openvswitch/vport.h
+++ b/net/openvswitch/vport.h
@@ -199,6 +199,7 @@ void ovs_vport_record_error(struct vport *, enum vport_err_type err_type);
 extern const struct vport_ops ovs_netdev_vport_ops;
 extern const struct vport_ops ovs_internal_vport_ops;
 extern const struct vport_ops ovs_gre_vport_ops;
+extern const struct vport_ops ovs_vxlan_vport_ops;
 
 static inline void ovs_skb_postpush_rcsum(struct sk_buff *skb,
 				      const void *start, unsigned int len)
-- 
1.7.1

Re: [PATCH net-next v3 6/6] openvswitch: Add vxlan tunneling support.

[Cong Wang]

On Wed, 24 Jul 2013 at 18:00 GMT, Pravin B Shelar <pshelar@nicira.com> wrote:
> +
> +config OPENVSWITCH_VXLAN
> +	bool "Open vSwitch VXLAN tunneling support"
> +	depends on INET
> +	depends on OPENVSWITCH
> +	depends on VXLAN && !(OPENVSWITCH=y && VXLAN=m)

This seems ugly, is it possible to make these vxlan code be a library as
iptunnel code?

Re: [PATCH net-next v3 6/6] openvswitch: Add vxlan tunneling support.

[Jesse Gross]

On Wed, Jul 24, 2013 at 11:44 PM, Cong Wang <xiyou.wangcong@gmail.com> wrote:
> On Wed, 24 Jul 2013 at 18:00 GMT, Pravin B Shelar <pshelar@nicira.com> wrote:
>> +
>> +config OPENVSWITCH_VXLAN
>> +     bool "Open vSwitch VXLAN tunneling support"
>> +     depends on INET
>> +     depends on OPENVSWITCH
>> +     depends on VXLAN && !(OPENVSWITCH=y && VXLAN=m)
>
> This seems ugly, is it possible to make these vxlan code be a library as
> iptunnel code?

The goal of separating things out this way is to provide nice layering
between the protocol code and the consumers. There needs to be some
interface between OVS and the tunnel code that straddles the boundary
and putting it in OVS makes the most sense from a hierarchy
perspective.