[ovs.git] / datapath / vport-lisp.c

/*
 * Copyright (c) 2011 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/version.h>

#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/route.h>
#include <net/udp.h>
#include <net/xfrm.h>

#include "datapath.h"
#include "gso.h"
#include "vport.h"

/*
 *  LISP encapsulation header:
 *
 *  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *  |N|L|E|V|I|flags|            Nonce/Map-Version                  |
 *  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *  |                 Instance ID/Locator Status Bits               |
 *  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 *
 */

/**
 * struct lisphdr - LISP header
 * @nonce_present: Flag indicating the presence of a 24 bit nonce value.
 * @locator_status_bits_present: Flag indicating the presence of Locator Status
 *                               Bits (LSB).
 * @solicit_echo_nonce: Flag indicating the use of the echo noncing mechanism.
 * @map_version_present: Flag indicating the use of mapping versioning.
 * @instance_id_present: Flag indicating the presence of a 24 bit Instance ID.
 * @reserved_flags: 3 bits reserved for future flags.
 * @nonce: 24 bit nonce value.
 * @map_version: 24 bit mapping version.
 * @locator_status_bits: Locator Status Bits: 32 bits when instance_id_present
 *                       is not set, 8 bits when it is.
 * @instance_id: 24 bit Instance ID
 */
struct lisphdr {
#ifdef __LITTLE_ENDIAN_BITFIELD
	__u8 reserved_flags:3;
	__u8 instance_id_present:1;
	__u8 map_version_present:1;
	__u8 solicit_echo_nonce:1;
	__u8 locator_status_bits_present:1;
	__u8 nonce_present:1;
#else
	__u8 nonce_present:1;
	__u8 locator_status_bits_present:1;
	__u8 solicit_echo_nonce:1;
	__u8 map_version_present:1;
	__u8 instance_id_present:1;
	__u8 reserved_flags:3;
#endif
	union {
		__u8 nonce[3];
		__u8 map_version[3];
	} u1;
	union {
		__be32 locator_status_bits;
		struct {
			__u8 instance_id[3];
			__u8 locator_status_bits;
		} word2;
	} u2;
};

#define LISP_HLEN (sizeof(struct udphdr) + sizeof(struct lisphdr))

/**
 * struct lisp_port - Keeps track of open UDP ports
 * @dst_port: lisp UDP port no.
 * @list: list element in @lisp_ports.
 * @lisp_rcv_socket: The socket created for this port number.
 * @name: vport name.
 */
struct lisp_port {
	__be16 dst_port;
	struct list_head list;
	struct socket *lisp_rcv_socket;
	char name[IFNAMSIZ];
};

static LIST_HEAD(lisp_ports);

static inline struct lisp_port *lisp_vport(const struct vport *vport)
{
	return vport_priv(vport);
}

static struct lisp_port *lisp_find_port(struct net *net, __be16 port)
{
	struct lisp_port *lisp_port;

	list_for_each_entry_rcu(lisp_port, &lisp_ports, list) {
		if (lisp_port->dst_port == port &&
			net_eq(sock_net(lisp_port->lisp_rcv_socket->sk), net))
			return lisp_port;
	}

	return NULL;
}

static inline struct lisphdr *lisp_hdr(const struct sk_buff *skb)
{
	return (struct lisphdr *)(udp_hdr(skb) + 1);
}

/* Convert 64 bit tunnel ID to 24 bit Instance ID. */
static void tunnel_id_to_instance_id(__be64 tun_id, __u8 *iid)
{

#ifdef __BIG_ENDIAN
	iid[0] = (__force __u8)(tun_id >> 16);
	iid[1] = (__force __u8)(tun_id >> 8);
	iid[2] = (__force __u8)tun_id;
#else
	iid[0] = (__force __u8)((__force u64)tun_id >> 40);
	iid[1] = (__force __u8)((__force u64)tun_id >> 48);
	iid[2] = (__force __u8)((__force u64)tun_id >> 56);
#endif
}

/* Convert 24 bit Instance ID to 64 bit tunnel ID. */
static __be64 instance_id_to_tunnel_id(__u8 *iid)
{
#ifdef __BIG_ENDIAN
	return (iid[0] << 16) | (iid[1] << 8) | iid[2];
#else
	return (__force __be64)(((__force u64)iid[0] << 40) |
				((__force u64)iid[1] << 48) |
				((__force u64)iid[2] << 56));
#endif
}

/* Compute source UDP port for outgoing packet.
 * Currently we use the flow hash.
 */
static u16 get_src_port(struct sk_buff *skb)
{
	int low;
	int high;
	unsigned int range;
	struct sw_flow_key *pkt_key = OVS_CB(skb)->pkt_key;
	u32 hash = jhash2((const u32 *)pkt_key,
			  sizeof(*pkt_key) / sizeof(u32), 0);

	inet_get_local_port_range(&low, &high);
	range = (high - low) + 1;
	return (((u64) hash * range) >> 32) + low;
}

static void lisp_build_header(const struct vport *vport,
			      struct sk_buff *skb)
{
	struct lisp_port *lisp_port = lisp_vport(vport);
	struct udphdr *udph = udp_hdr(skb);
	struct lisphdr *lisph = (struct lisphdr *)(udph + 1);
	const struct ovs_key_ipv4_tunnel *tun_key = OVS_CB(skb)->tun_key;

	udph->dest = lisp_port->dst_port;
	udph->source = htons(get_src_port(skb));
	udph->check = 0;
	udph->len = htons(skb->len - skb_transport_offset(skb));

	lisph->nonce_present = 0;	/* We don't support echo nonce algorithm */
	lisph->locator_status_bits_present = 1;	/* Set LSB */
	lisph->solicit_echo_nonce = 0;	/* No echo noncing */
	lisph->map_version_present = 0;	/* No mapping versioning, nonce instead */
	lisph->instance_id_present = 1;	/* Store the tun_id as Instance ID  */
	lisph->reserved_flags = 0;	/* Reserved flags, set to 0  */

	lisph->u1.nonce[0] = 0;
	lisph->u1.nonce[1] = 0;
	lisph->u1.nonce[2] = 0;

	tunnel_id_to_instance_id(tun_key->tun_id, &lisph->u2.word2.instance_id[0]);
	lisph->u2.word2.locator_status_bits = 1;
}

/* Called with rcu_read_lock and BH disabled. */
static int lisp_rcv(struct sock *sk, struct sk_buff *skb)
{
	struct lisp_port *lisp_port;
	struct lisphdr *lisph;
	struct iphdr *iph, *inner_iph;
	struct ovs_key_ipv4_tunnel tun_key;
	__be64 key;
	struct ethhdr *ethh;
	__be16 protocol;

	lisp_port = lisp_find_port(dev_net(skb->dev), udp_hdr(skb)->dest);
	if (unlikely(!lisp_port))
		goto error;

	if (iptunnel_pull_header(skb, LISP_HLEN, 0))
		goto error;

	lisph = lisp_hdr(skb);

	if (lisph->instance_id_present != 1)
		key = 0;
	else
		key = instance_id_to_tunnel_id(&lisph->u2.word2.instance_id[0]);

	/* Save outer tunnel values */
	iph = ip_hdr(skb);
	ovs_flow_tun_key_init(&tun_key, iph, key, TUNNEL_KEY);

	/* Drop non-IP inner packets */
	inner_iph = (struct iphdr *)(lisph + 1);
	switch (inner_iph->version) {
	case 4:
		protocol = htons(ETH_P_IP);
		break;
	case 6:
		protocol = htons(ETH_P_IPV6);
		break;
	default:
		goto error;
	}
	skb->protocol = protocol;

	/* Add Ethernet header */
	ethh = (struct ethhdr *)skb_push(skb, ETH_HLEN);
	memset(ethh, 0, ETH_HLEN);
	ethh->h_dest[0] = 0x02;
	ethh->h_source[0] = 0x02;
	ethh->h_proto = protocol;

	ovs_skb_postpush_rcsum(skb, skb->data, ETH_HLEN);

	ovs_vport_receive(vport_from_priv(lisp_port), skb, &tun_key);
	goto out;

error:
	kfree_skb(skb);
out:
	return 0;
}

/* Arbitrary value.  Irrelevant as long as it's not 0 since we set the handler. */
#define UDP_ENCAP_LISP 1
static int lisp_socket_init(struct lisp_port *lisp_port, struct net *net)
{
	struct sockaddr_in sin;
	int err;

	err = sock_create_kern(AF_INET, SOCK_DGRAM, 0,
			       &lisp_port->lisp_rcv_socket);
	if (err)
		goto error;

	/* release net ref. */
	sk_change_net(lisp_port->lisp_rcv_socket->sk, net);

	sin.sin_family = AF_INET;
	sin.sin_addr.s_addr = htonl(INADDR_ANY);
	sin.sin_port = lisp_port->dst_port;

	err = kernel_bind(lisp_port->lisp_rcv_socket, (struct sockaddr *)&sin,
			  sizeof(struct sockaddr_in));
	if (err)
		goto error_sock;

	udp_sk(lisp_port->lisp_rcv_socket->sk)->encap_type = UDP_ENCAP_LISP;
	udp_sk(lisp_port->lisp_rcv_socket->sk)->encap_rcv = lisp_rcv;

	udp_encap_enable();

	return 0;

error_sock:
	sk_release_kernel(lisp_port->lisp_rcv_socket->sk);
error:
	pr_warn("cannot register lisp protocol handler: %d\n", err);
	return err;
}

static int lisp_get_options(const struct vport *vport, struct sk_buff *skb)
{
	struct lisp_port *lisp_port = lisp_vport(vport);

	if (nla_put_u16(skb, OVS_TUNNEL_ATTR_DST_PORT, ntohs(lisp_port->dst_port)))
		return -EMSGSIZE;
	return 0;
}

static void lisp_tnl_destroy(struct vport *vport)
{
	struct lisp_port *lisp_port = lisp_vport(vport);

	list_del_rcu(&lisp_port->list);
	/* Release socket */
	sk_release_kernel(lisp_port->lisp_rcv_socket->sk);

	ovs_vport_deferred_free(vport);
}

static struct vport *lisp_tnl_create(const struct vport_parms *parms)
{
	struct net *net = ovs_dp_get_net(parms->dp);
	struct nlattr *options = parms->options;
	struct lisp_port *lisp_port;
	struct vport *vport;
	struct nlattr *a;
	int err;
	u16 dst_port;

	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;
	}

	/* Verify if we already have a socket created for this port */
	if (lisp_find_port(net, htons(dst_port))) {
		err = -EEXIST;
		goto error;
	}

	vport = ovs_vport_alloc(sizeof(struct lisp_port),
				&ovs_lisp_vport_ops, parms);
	if (IS_ERR(vport))
		return vport;

	lisp_port = lisp_vport(vport);
	lisp_port->dst_port = htons(dst_port);
	strncpy(lisp_port->name, parms->name, IFNAMSIZ);

	err = lisp_socket_init(lisp_port, net);
	if (err)
		goto error_free;

	list_add_tail_rcu(&lisp_port->list, &lisp_ports);
	return vport;

error_free:
	ovs_vport_free(vport);
error:
	return ERR_PTR(err);
}

static void lisp_fix_segment(struct sk_buff *skb)
{
	struct udphdr *udph = udp_hdr(skb);

	udph->len = htons(skb->len - skb_transport_offset(skb));
}

static void handle_offloads(struct sk_buff *skb)
{
	if (skb_is_gso(skb))
		OVS_GSO_CB(skb)->fix_segment = lisp_fix_segment;
	else if (skb->ip_summed != CHECKSUM_PARTIAL)
		skb->ip_summed = CHECKSUM_NONE;
}

static int lisp_send(struct vport *vport, struct sk_buff *skb)
{
	int network_offset = skb_network_offset(skb);
	struct rtable *rt;
	int min_headroom;
	__be32 saddr;
	__be16 df;
	int sent_len;
	int err;

	if (unlikely(!OVS_CB(skb)->tun_key))
		return -EINVAL;

	if (skb->protocol != htons(ETH_P_IP) &&
	    skb->protocol != htons(ETH_P_IPV6)) {
		kfree_skb(skb);
		return 0;
	}

	/* Route lookup */
	saddr = OVS_CB(skb)->tun_key->ipv4_src;
	rt = find_route(ovs_dp_get_net(vport->dp),
			&saddr,
			OVS_CB(skb)->tun_key->ipv4_dst,
			IPPROTO_UDP,
			OVS_CB(skb)->tun_key->ipv4_tos,
			skb->mark);
	if (IS_ERR(rt)) {
		err = PTR_ERR(rt);
		goto error;
	}

	min_headroom = LL_RESERVED_SPACE(rt_dst(rt).dev) + rt_dst(rt).header_len
			+ sizeof(struct iphdr) + LISP_HLEN;

	if (skb_headroom(skb) < min_headroom || skb_header_cloned(skb)) {
		int head_delta = SKB_DATA_ALIGN(min_headroom -
						skb_headroom(skb) +
						16);

		err = pskb_expand_head(skb, max_t(int, head_delta, 0),
					0, GFP_ATOMIC);
		if (unlikely(err))
			goto err_free_rt;
	}

	/* Reset l2 headers. */
	skb_pull(skb, network_offset);
	skb_reset_mac_header(skb);
	vlan_set_tci(skb, 0);

	skb_reset_inner_headers(skb);

	__skb_push(skb, LISP_HLEN);
	skb_reset_transport_header(skb);

	lisp_build_header(vport, skb);

	/* Offloading */
	handle_offloads(skb);
	skb->local_df = 1;

	df = OVS_CB(skb)->tun_key->tun_flags &
				  TUNNEL_DONT_FRAGMENT ?  htons(IP_DF) : 0;
	sent_len = iptunnel_xmit(rt, skb,
			     saddr, OVS_CB(skb)->tun_key->ipv4_dst,
			     IPPROTO_UDP, OVS_CB(skb)->tun_key->ipv4_tos,
			     OVS_CB(skb)->tun_key->ipv4_ttl, df);

	return sent_len > 0 ? sent_len + network_offset : sent_len;

err_free_rt:
	ip_rt_put(rt);
error:
	return err;
}

static const char *lisp_get_name(const struct vport *vport)
{
	struct lisp_port *lisp_port = lisp_vport(vport);
	return lisp_port->name;
}

const struct vport_ops ovs_lisp_vport_ops = {
	.type		= OVS_VPORT_TYPE_LISP,
	.create		= lisp_tnl_create,
	.destroy	= lisp_tnl_destroy,
	.get_name	= lisp_get_name,
	.get_options	= lisp_get_options,
	.send		= lisp_send,
};
Commit	Line	Data
a6ae068b LJ	1	/*
	2	* Copyright (c) 2011 Nicira, Inc.
	3	* Copyright (c) 2013 Cisco Systems, Inc.
	4	*
	5	* This program is free software; you can redistribute it and/or
	6	* modify it under the terms of version 2 of the GNU General Public
	7	* License as published by the Free Software Foundation.
	8	*
	9	* This program is distributed in the hope that it will be useful, but
	10	* WITHOUT ANY WARRANTY; without even the implied warranty of
	11	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	12	* General Public License for more details.
	13	*
	14	* You should have received a copy of the GNU General Public License
	15	* along with this program; if not, write to the Free Software
	16	* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
	17	* 02110-1301, USA
	18	*/
	19
	20	#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
	21
	22	#include <linux/version.h>
a6ae068b LJ	23
	24	#include <linux/in.h>
	25	#include <linux/ip.h>
a6ae068b	26	#include <linux/net.h>
85c9de19	27	#include <linux/rculist.h>
a6ae068b LJ	28	#include <linux/udp.h>
	29
	30	#include <net/icmp.h>
	31	#include <net/ip.h>
11aa8dff	32	#include <net/route.h>
a6ae068b	33	#include <net/udp.h>
11aa8dff	34	#include <net/xfrm.h>
a6ae068b LJ	35
a6ae068b LJ	36	#include "datapath.h"
3cf54a56	37	#include "gso.h"
a6ae068b LJ	38	#include "vport.h"
a6ae068b LJ	39
a6ae068b LJ	40	/*
	41	* LISP encapsulation header:
	42	*
	43	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	44	* \|N\|L\|E\|V\|I\|flags\| Nonce/Map-Version \|
	45	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	46	* \| Instance ID/Locator Status Bits \|
	47	* +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
	48	*
	49	*/
	50
	51	/**
	52	* struct lisphdr - LISP header
	53	* @nonce_present: Flag indicating the presence of a 24 bit nonce value.
	54	* @locator_status_bits_present: Flag indicating the presence of Locator Status
	55	* Bits (LSB).
	56	* @solicit_echo_nonce: Flag indicating the use of the echo noncing mechanism.
	57	* @map_version_present: Flag indicating the use of mapping versioning.
	58	* @instance_id_present: Flag indicating the presence of a 24 bit Instance ID.
	59	* @reserved_flags: 3 bits reserved for future flags.
	60	* @nonce: 24 bit nonce value.
	61	* @map_version: 24 bit mapping version.
	62	* @locator_status_bits: Locator Status Bits: 32 bits when instance_id_present
	63	* is not set, 8 bits when it is.
	64	* @instance_id: 24 bit Instance ID
	65	*/
	66	struct lisphdr {
	67	#ifdef __LITTLE_ENDIAN_BITFIELD
	68	__u8 reserved_flags:3;
	69	__u8 instance_id_present:1;
	70	__u8 map_version_present:1;
	71	__u8 solicit_echo_nonce:1;
	72	__u8 locator_status_bits_present:1;
	73	__u8 nonce_present:1;
	74	#else
	75	__u8 nonce_present:1;
	76	__u8 locator_status_bits_present:1;
	77	__u8 solicit_echo_nonce:1;
	78	__u8 map_version_present:1;
	79	__u8 instance_id_present:1;
	80	__u8 reserved_flags:3;
	81	#endif
	82	union {
	83	__u8 nonce[3];
	84	__u8 map_version[3];
	85	} u1;
	86	union {
	87	__be32 locator_status_bits;
	88	struct {
	89	__u8 instance_id[3];
	90	__u8 locator_status_bits;
	91	} word2;
	92	} u2;
	93	};
	94
	95	#define LISP_HLEN (sizeof(struct udphdr) + sizeof(struct lisphdr))
	96
a6ae068b LJ	97	/**
a6ae068b LJ	98	* struct lisp_port - Keeps track of open UDP ports
c405d282 PS	99	* @dst_port: lisp UDP port no.
	100	* @list: list element in @lisp_ports.
	101	* @lisp_rcv_socket: The socket created for this port number.
	102	* @name: vport name.
a6ae068b LJ	103	*/
a6ae068b LJ	104	struct lisp_port {
c405d282	105	__be16 dst_port;
a6ae068b	106	struct list_head list;
a6ae068b	107	struct socket *lisp_rcv_socket;
c405d282	108	char name[IFNAMSIZ];
a6ae068b LJ	109	};
	110
	111	static LIST_HEAD(lisp_ports);
	112
c405d282 PS	113	static inline struct lisp_port lisp_vport(const struct vport vport)
	114	{
	115	return vport_priv(vport);
	116	}
	117
85c9de19	118	static struct lisp_port lisp_find_port(struct net net, __be16 port)
a6ae068b LJ	119	{
	120	struct lisp_port *lisp_port;
	121
85c9de19	122	list_for_each_entry_rcu(lisp_port, &lisp_ports, list) {
c405d282	123	if (lisp_port->dst_port == port &&
a6ae068b LJ	124	net_eq(sock_net(lisp_port->lisp_rcv_socket->sk), net))
	125	return lisp_port;
	126	}
	127
	128	return NULL;
	129	}
	130
	131	static inline struct lisphdr lisp_hdr(const struct sk_buff skb)
	132	{
	133	return (struct lisphdr *)(udp_hdr(skb) + 1);
	134	}
	135
a6ae068b LJ	136	/* Convert 64 bit tunnel ID to 24 bit Instance ID. */
	137	static void tunnel_id_to_instance_id(__be64 tun_id, __u8 *iid)
	138	{
	139
	140	#ifdef __BIG_ENDIAN
	141	iid[0] = (__force __u8)(tun_id >> 16);
	142	iid[1] = (__force __u8)(tun_id >> 8);
	143	iid[2] = (__force __u8)tun_id;
	144	#else
	145	iid[0] = (__force __u8)((__force u64)tun_id >> 40);
	146	iid[1] = (__force __u8)((__force u64)tun_id >> 48);
	147	iid[2] = (__force __u8)((__force u64)tun_id >> 56);
	148	#endif
	149	}
	150
	151	/* Convert 24 bit Instance ID to 64 bit tunnel ID. */
	152	static __be64 instance_id_to_tunnel_id(__u8 *iid)
	153	{
	154	#ifdef __BIG_ENDIAN
	155	return (iid[0] << 16) \| (iid[1] << 8) \| iid[2];
	156	#else
	157	return (__force __be64)(((__force u64)iid[0] << 40) \|
	158	((__force u64)iid[1] << 48) \|
	159	((__force u64)iid[2] << 56));
	160	#endif
	161	}
	162
11aa8dff PS	163	/* Compute source UDP port for outgoing packet.
	164	* Currently we use the flow hash.
	165	*/
3cf54a56	166	static u16 get_src_port(struct sk_buff *skb)
11aa8dff PS	167	{
	168	int low;
	169	int high;
	170	unsigned int range;
	171	struct sw_flow_key *pkt_key = OVS_CB(skb)->pkt_key;
	172	u32 hash = jhash2((const u32 *)pkt_key,
	173	sizeof(*pkt_key) / sizeof(u32), 0);
	174
	175	inet_get_local_port_range(&low, &high);
	176	range = (high - low) + 1;
	177	return (((u64) hash * range) >> 32) + low;
	178	}
	179
85c9de19	180	static void lisp_build_header(const struct vport *vport,
3cf54a56	181	struct sk_buff *skb)
a6ae068b	182	{
c405d282	183	struct lisp_port *lisp_port = lisp_vport(vport);
a6ae068b LJ	184	struct udphdr *udph = udp_hdr(skb);
	185	struct lisphdr lisph = (struct lisphdr )(udph + 1);
	186	const struct ovs_key_ipv4_tunnel *tun_key = OVS_CB(skb)->tun_key;
a6ae068b	187
c405d282	188	udph->dest = lisp_port->dst_port;
3cf54a56	189	udph->source = htons(get_src_port(skb));
a6ae068b LJ	190	udph->check = 0;
	191	udph->len = htons(skb->len - skb_transport_offset(skb));
	192
	193	lisph->nonce_present = 0; /* We don't support echo nonce algorithm */
	194	lisph->locator_status_bits_present = 1; /* Set LSB */
	195	lisph->solicit_echo_nonce = 0; /* No echo noncing */
	196	lisph->map_version_present = 0; /* No mapping versioning, nonce instead */
	197	lisph->instance_id_present = 1; /* Store the tun_id as Instance ID */
	198	lisph->reserved_flags = 0; /* Reserved flags, set to 0 */
	199
	200	lisph->u1.nonce[0] = 0;
	201	lisph->u1.nonce[1] = 0;
	202	lisph->u1.nonce[2] = 0;
	203
85c9de19	204	tunnel_id_to_instance_id(tun_key->tun_id, &lisph->u2.word2.instance_id[0]);
a6ae068b	205	lisph->u2.word2.locator_status_bits = 1;
a6ae068b LJ	206	}
	207
	208	/* Called with rcu_read_lock and BH disabled. */
	209	static int lisp_rcv(struct sock sk, struct sk_buff skb)
	210	{
85c9de19	211	struct lisp_port *lisp_port;
a6ae068b	212	struct lisphdr *lisph;
a6ae068b LJ	213	struct iphdr iph, inner_iph;
	214	struct ovs_key_ipv4_tunnel tun_key;
	215	__be64 key;
a6ae068b LJ	216	struct ethhdr *ethh;
	217	__be16 protocol;
	218
85c9de19 PS	219	lisp_port = lisp_find_port(dev_net(skb->dev), udp_hdr(skb)->dest);
	220	if (unlikely(!lisp_port))
	221	goto error;
	222
971fd4ab	223	if (iptunnel_pull_header(skb, LISP_HLEN, 0))
a6ae068b LJ	224	goto error;
	225
	226	lisph = lisp_hdr(skb);
	227
a6ae068b LJ	228	if (lisph->instance_id_present != 1)
	229	key = 0;
	230	else
	231	key = instance_id_to_tunnel_id(&lisph->u2.word2.instance_id[0]);
	232
a6ae068b	233	/* Save outer tunnel values */
85c9de19	234	iph = ip_hdr(skb);
adda018c	235	ovs_flow_tun_key_init(&tun_key, iph, key, TUNNEL_KEY);
a6ae068b LJ	236
	237	/* Drop non-IP inner packets */
	238	inner_iph = (struct iphdr *)(lisph + 1);
	239	switch (inner_iph->version) {
	240	case 4:
	241	protocol = htons(ETH_P_IP);
	242	break;
	243	case 6:
	244	protocol = htons(ETH_P_IPV6);
	245	break;
	246	default:
	247	goto error;
	248	}
971fd4ab	249	skb->protocol = protocol;
a6ae068b LJ	250
	251	/* Add Ethernet header */
	252	ethh = (struct ethhdr *)skb_push(skb, ETH_HLEN);
	253	memset(ethh, 0, ETH_HLEN);
	254	ethh->h_dest[0] = 0x02;
	255	ethh->h_source[0] = 0x02;
	256	ethh->h_proto = protocol;
	257
3cfede14 PS	258	ovs_skb_postpush_rcsum(skb, skb->data, ETH_HLEN);
3cfede14 PS	259
971fd4ab	260	ovs_vport_receive(vport_from_priv(lisp_port), skb, &tun_key);
a6ae068b LJ	261	goto out;
	262
	263	error:
	264	kfree_skb(skb);
	265	out:
	266	return 0;
	267	}
	268
	269	/* Arbitrary value. Irrelevant as long as it's not 0 since we set the handler. */
	270	#define UDP_ENCAP_LISP 1
	271	static int lisp_socket_init(struct lisp_port lisp_port, struct net net)
	272	{
a6ae068b	273	struct sockaddr_in sin;
c405d282	274	int err;
a6ae068b LJ	275
	276	err = sock_create_kern(AF_INET, SOCK_DGRAM, 0,
	277	&lisp_port->lisp_rcv_socket);
	278	if (err)
	279	goto error;
	280
	281	/* release net ref. */
	282	sk_change_net(lisp_port->lisp_rcv_socket->sk, net);
	283
	284	sin.sin_family = AF_INET;
	285	sin.sin_addr.s_addr = htonl(INADDR_ANY);
c405d282	286	sin.sin_port = lisp_port->dst_port;
a6ae068b LJ	287
	288	err = kernel_bind(lisp_port->lisp_rcv_socket, (struct sockaddr *)&sin,
	289	sizeof(struct sockaddr_in));
	290	if (err)
	291	goto error_sock;
	292
	293	udp_sk(lisp_port->lisp_rcv_socket->sk)->encap_type = UDP_ENCAP_LISP;
	294	udp_sk(lisp_port->lisp_rcv_socket->sk)->encap_rcv = lisp_rcv;
	295
	296	udp_encap_enable();
	297
	298	return 0;
	299
	300	error_sock:
	301	sk_release_kernel(lisp_port->lisp_rcv_socket->sk);
	302	error:
	303	pr_warn("cannot register lisp protocol handler: %d\n", err);
	304	return err;
	305	}
	306
c405d282	307	static int lisp_get_options(const struct vport vport, struct sk_buff skb)
a6ae068b	308	{
c405d282	309	struct lisp_port *lisp_port = lisp_vport(vport);
a6ae068b	310
c405d282 PS	311	if (nla_put_u16(skb, OVS_TUNNEL_ATTR_DST_PORT, ntohs(lisp_port->dst_port)))
	312	return -EMSGSIZE;
	313	return 0;
a6ae068b LJ	314	}
a6ae068b LJ	315
c405d282	316	static void lisp_tnl_destroy(struct vport *vport)
a6ae068b	317	{
c405d282 PS	318	struct lisp_port *lisp_port = lisp_vport(vport);
c405d282 PS	319
85c9de19 PS	320	list_del_rcu(&lisp_port->list);
	321	/* Release socket */
	322	sk_release_kernel(lisp_port->lisp_rcv_socket->sk);
c405d282 PS	323
c405d282 PS	324	ovs_vport_deferred_free(vport);
85c9de19 PS	325	}
85c9de19 PS	326
c405d282	327	static struct vport lisp_tnl_create(const struct vport_parms parms)
85c9de19	328	{
c405d282 PS	329	struct net *net = ovs_dp_get_net(parms->dp);
c405d282 PS	330	struct nlattr *options = parms->options;
85c9de19	331	struct lisp_port *lisp_port;
c405d282	332	struct vport *vport;
a6ae068b LJ	333	struct nlattr *a;
	334	int err;
	335	u16 dst_port;
a6ae068b LJ	336
	337	if (!options) {
	338	err = -EINVAL;
c405d282	339	goto error;
a6ae068b LJ	340	}
	341
	342	a = nla_find_nested(options, OVS_TUNNEL_ATTR_DST_PORT);
	343	if (a && nla_len(a) == sizeof(u16)) {
	344	dst_port = nla_get_u16(a);
	345	} else {
	346	/* Require destination port from userspace. */
	347	err = -EINVAL;
c405d282	348	goto error;
a6ae068b LJ	349	}
	350
	351	/* Verify if we already have a socket created for this port */
c405d282	352	if (lisp_find_port(net, htons(dst_port))) {
85c9de19	353	err = -EEXIST;
c405d282	354	goto error;
a6ae068b LJ	355	}
a6ae068b LJ	356
c405d282 PS	357	vport = ovs_vport_alloc(sizeof(struct lisp_port),
	358	&ovs_lisp_vport_ops, parms);
	359	if (IS_ERR(vport))
	360	return vport;
a6ae068b	361
c405d282 PS	362	lisp_port = lisp_vport(vport);
	363	lisp_port->dst_port = htons(dst_port);
	364	strncpy(lisp_port->name, parms->name, IFNAMSIZ);
a6ae068b LJ	365
	366	err = lisp_socket_init(lisp_port, net);
	367	if (err)
c405d282	368	goto error_free;
a6ae068b	369
c405d282 PS	370	list_add_tail_rcu(&lisp_port->list, &lisp_ports);
c405d282 PS	371	return vport;
a6ae068b	372
c405d282 PS	373	error_free:
c405d282 PS	374	ovs_vport_free(vport);
a6ae068b	375	error:
c405d282	376	return ERR_PTR(err);
a6ae068b LJ	377	}
a6ae068b LJ	378
3cf54a56	379	static void lisp_fix_segment(struct sk_buff *skb)
11aa8dff	380	{
3cf54a56	381	struct udphdr *udph = udp_hdr(skb);
11aa8dff	382
3cf54a56	383	udph->len = htons(skb->len - skb_transport_offset(skb));
11aa8dff PS	384	}
11aa8dff PS	385
3cf54a56	386	static void handle_offloads(struct sk_buff *skb)
11aa8dff	387	{
3cf54a56 PS	388	if (skb_is_gso(skb))
	389	OVS_GSO_CB(skb)->fix_segment = lisp_fix_segment;
	390	else if (skb->ip_summed != CHECKSUM_PARTIAL)
	391	skb->ip_summed = CHECKSUM_NONE;
11aa8dff PS	392	}
11aa8dff PS	393
3cf54a56	394	static int lisp_send(struct vport vport, struct sk_buff skb)
11aa8dff	395	{
3cf54a56	396	int network_offset = skb_network_offset(skb);
11aa8dff	397	struct rtable *rt;
3cf54a56	398	int min_headroom;
11aa8dff	399	__be32 saddr;
3cf54a56 PS	400	__be16 df;
3cf54a56 PS	401	int sent_len;
11aa8dff	402	int err;
3cf54a56 PS	403
	404	if (unlikely(!OVS_CB(skb)->tun_key))
	405	return -EINVAL;
	406
	407	if (skb->protocol != htons(ETH_P_IP) &&
	408	skb->protocol != htons(ETH_P_IPV6)) {
	409	kfree_skb(skb);
	410	return 0;
	411	}
11aa8dff PS	412
	413	/* Route lookup */
	414	saddr = OVS_CB(skb)->tun_key->ipv4_src;
	415	rt = find_route(ovs_dp_get_net(vport->dp),
	416	&saddr,
	417	OVS_CB(skb)->tun_key->ipv4_dst,
3cf54a56	418	IPPROTO_UDP,
11aa8dff	419	OVS_CB(skb)->tun_key->ipv4_tos,
3025a772	420	skb->mark);
11aa8dff PS	421	if (IS_ERR(rt)) {
	422	err = PTR_ERR(rt);
	423	goto error;
	424	}
	425
11aa8dff	426	min_headroom = LL_RESERVED_SPACE(rt_dst(rt).dev) + rt_dst(rt).header_len
3cf54a56	427	+ sizeof(struct iphdr) + LISP_HLEN;
11aa8dff PS	428
	429	if (skb_headroom(skb) < min_headroom \|\| skb_header_cloned(skb)) {
	430	int head_delta = SKB_DATA_ALIGN(min_headroom -
	431	skb_headroom(skb) +
	432	16);
	433
	434	err = pskb_expand_head(skb, max_t(int, head_delta, 0),
	435	0, GFP_ATOMIC);
	436	if (unlikely(err))
	437	goto err_free_rt;
	438	}
	439
df7d2c59	440	/* Reset l2 headers. */
3cf54a56 PS	441	skb_pull(skb, network_offset);
3cf54a56 PS	442	skb_reset_mac_header(skb);
df7d2c59 PS	443	vlan_set_tci(skb, 0);
df7d2c59 PS	444
3cf54a56 PS	445	skb_reset_inner_headers(skb);
	446
	447	__skb_push(skb, LISP_HLEN);
	448	skb_reset_transport_header(skb);
	449
	450	lisp_build_header(vport, skb);
	451
11aa8dff	452	/* Offloading */
3cf54a56 PS	453	handle_offloads(skb);
	454	skb->local_df = 1;
	455
	456	df = OVS_CB(skb)->tun_key->tun_flags &
11aa8dff	457	TUNNEL_DONT_FRAGMENT ? htons(IP_DF) : 0;
2b897011	458	sent_len = iptunnel_xmit(rt, skb,
3cf54a56 PS	459	saddr, OVS_CB(skb)->tun_key->ipv4_dst,
	460	IPPROTO_UDP, OVS_CB(skb)->tun_key->ipv4_tos,
	461	OVS_CB(skb)->tun_key->ipv4_ttl, df);
11aa8dff	462
3cf54a56	463	return sent_len > 0 ? sent_len + network_offset : sent_len;
11aa8dff PS	464
	465	err_free_rt:
	466	ip_rt_put(rt);
	467	error:
	468	return err;
	469	}
	470
c405d282	471	static const char lisp_get_name(const struct vport vport)
a6ae068b	472	{
c405d282 PS	473	struct lisp_port *lisp_port = lisp_vport(vport);
c405d282 PS	474	return lisp_port->name;
a6ae068b LJ	475	}
	476
	477	const struct vport_ops ovs_lisp_vport_ops = {
	478	.type = OVS_VPORT_TYPE_LISP,
a6ae068b LJ	479	.create = lisp_tnl_create,
a6ae068b LJ	480	.destroy = lisp_tnl_destroy,
c405d282	481	.get_name = lisp_get_name,
85c9de19	482	.get_options = lisp_get_options,
3cf54a56	483	.send = lisp_send,
a6ae068b	484	};