[mirror_ubuntu-artful-kernel.git] / net / ipv4 / geneve.c

/*
 * Geneve: Generic Network Virtualization Encapsulation
 *
 * Copyright (c) 2014 Nicira, Inc.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * as published by the Free Software Foundation; either version
 * 2 of the License, or (at your option) any later version.
 */

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/module.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/skbuff.h>
#include <linux/list.h>
#include <linux/netdevice.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/udp.h>
#include <linux/igmp.h>
#include <linux/etherdevice.h>
#include <linux/if_ether.h>
#include <linux/if_vlan.h>
#include <linux/ethtool.h>
#include <linux/mutex.h>
#include <net/arp.h>
#include <net/ndisc.h>
#include <net/ip.h>
#include <net/ip_tunnels.h>
#include <net/icmp.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/geneve.h>
#include <net/protocol.h>
#include <net/udp_tunnel.h>
#if IS_ENABLED(CONFIG_IPV6)
#include <net/ipv6.h>
#include <net/addrconf.h>
#include <net/ip6_tunnel.h>
#include <net/ip6_checksum.h>
#endif

/* Protects sock_list and refcounts. */
static DEFINE_MUTEX(geneve_mutex);

/* per-network namespace private data for this module */
struct geneve_net {
	struct list_head	sock_list;
};

static int geneve_net_id;

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

static struct geneve_sock *geneve_find_sock(struct net *net,
					    sa_family_t family, __be16 port)
{
	struct geneve_net *gn = net_generic(net, geneve_net_id);
	struct geneve_sock *gs;

	list_for_each_entry(gs, &gn->sock_list, list) {
		if (inet_sk(gs->sock->sk)->inet_sport == port &&
		    inet_sk(gs->sock->sk)->sk.sk_family == family)
			return gs;
	}

	return NULL;
}

static void geneve_build_header(struct genevehdr *geneveh,
				__be16 tun_flags, u8 vni[3],
				u8 options_len, u8 *options)
{
	geneveh->ver = GENEVE_VER;
	geneveh->opt_len = options_len / 4;
	geneveh->oam = !!(tun_flags & TUNNEL_OAM);
	geneveh->critical = !!(tun_flags & TUNNEL_CRIT_OPT);
	geneveh->rsvd1 = 0;
	memcpy(geneveh->vni, vni, 3);
	geneveh->proto_type = htons(ETH_P_TEB);
	geneveh->rsvd2 = 0;

	memcpy(geneveh->options, options, options_len);
}

/* Transmit a fully formatted Geneve frame.
 *
 * When calling this function. The skb->data should point
 * to the geneve header which is fully formed.
 *
 * This function will add other UDP tunnel headers.
 */
int geneve_xmit_skb(struct geneve_sock *gs, struct rtable *rt,
		    struct sk_buff *skb, __be32 src, __be32 dst, __u8 tos,
		    __u8 ttl, __be16 df, __be16 src_port, __be16 dst_port,
		    __be16 tun_flags, u8 vni[3], u8 opt_len, u8 *opt,
		    bool xnet)
{
	struct genevehdr *gnvh;
	int min_headroom;
	int err;

	skb = udp_tunnel_handle_offloads(skb, !gs->sock->sk->sk_no_check_tx);
	if (IS_ERR(skb))
		return PTR_ERR(skb);

	min_headroom = LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len
			+ GENEVE_BASE_HLEN + opt_len + sizeof(struct iphdr)
			+ (skb_vlan_tag_present(skb) ? VLAN_HLEN : 0);

	err = skb_cow_head(skb, min_headroom);
	if (unlikely(err)) {
		kfree_skb(skb);
		return err;
	}

	skb = vlan_hwaccel_push_inside(skb);
	if (unlikely(!skb))
		return -ENOMEM;

	gnvh = (struct genevehdr *)__skb_push(skb, sizeof(*gnvh) + opt_len);
	geneve_build_header(gnvh, tun_flags, vni, opt_len, opt);

	skb_set_inner_protocol(skb, htons(ETH_P_TEB));

	return udp_tunnel_xmit_skb(gs->sock, rt, skb, src, dst,
				   tos, ttl, df, src_port, dst_port, xnet);
}
EXPORT_SYMBOL_GPL(geneve_xmit_skb);

static int geneve_hlen(struct genevehdr *gh)
{
	return sizeof(*gh) + gh->opt_len * 4;
}

static struct sk_buff **geneve_gro_receive(struct sk_buff **head,
					   struct sk_buff *skb,
					   struct udp_offload *uoff)
{
	struct sk_buff *p, **pp = NULL;
	struct genevehdr *gh, *gh2;
	unsigned int hlen, gh_len, off_gnv;
	const struct packet_offload *ptype;
	__be16 type;
	int flush = 1;

	off_gnv = skb_gro_offset(skb);
	hlen = off_gnv + sizeof(*gh);
	gh = skb_gro_header_fast(skb, off_gnv);
	if (skb_gro_header_hard(skb, hlen)) {
		gh = skb_gro_header_slow(skb, hlen, off_gnv);
		if (unlikely(!gh))
			goto out;
	}

	if (gh->ver != GENEVE_VER || gh->oam)
		goto out;
	gh_len = geneve_hlen(gh);

	hlen = off_gnv + gh_len;
	if (skb_gro_header_hard(skb, hlen)) {
		gh = skb_gro_header_slow(skb, hlen, off_gnv);
		if (unlikely(!gh))
			goto out;
	}

	flush = 0;

	for (p = *head; p; p = p->next) {
		if (!NAPI_GRO_CB(p)->same_flow)
			continue;

		gh2 = (struct genevehdr *)(p->data + off_gnv);
		if (gh->opt_len != gh2->opt_len ||
		    memcmp(gh, gh2, gh_len)) {
			NAPI_GRO_CB(p)->same_flow = 0;
			continue;
		}
	}

	type = gh->proto_type;

	rcu_read_lock();
	ptype = gro_find_receive_by_type(type);
	if (ptype == NULL) {
		flush = 1;
		goto out_unlock;
	}

	skb_gro_pull(skb, gh_len);
	skb_gro_postpull_rcsum(skb, gh, gh_len);
	pp = ptype->callbacks.gro_receive(head, skb);

out_unlock:
	rcu_read_unlock();
out:
	NAPI_GRO_CB(skb)->flush |= flush;

	return pp;
}

static int geneve_gro_complete(struct sk_buff *skb, int nhoff,
			       struct udp_offload *uoff)
{
	struct genevehdr *gh;
	struct packet_offload *ptype;
	__be16 type;
	int gh_len;
	int err = -ENOSYS;

	udp_tunnel_gro_complete(skb, nhoff);

	gh = (struct genevehdr *)(skb->data + nhoff);
	gh_len = geneve_hlen(gh);
	type = gh->proto_type;

	rcu_read_lock();
	ptype = gro_find_complete_by_type(type);
	if (ptype != NULL)
		err = ptype->callbacks.gro_complete(skb, nhoff + gh_len);

	rcu_read_unlock();
	return err;
}

static void geneve_notify_add_rx_port(struct geneve_sock *gs)
{
	struct sock *sk = gs->sock->sk;
	sa_family_t sa_family = sk->sk_family;
	int err;

	if (sa_family == AF_INET) {
		err = udp_add_offload(&gs->udp_offloads);
		if (err)
			pr_warn("geneve: udp_add_offload failed with status %d\n",
				err);
	}
}

static void geneve_notify_del_rx_port(struct geneve_sock *gs)
{
	struct sock *sk = gs->sock->sk;
	sa_family_t sa_family = sk->sk_family;

	if (sa_family == AF_INET)
		udp_del_offload(&gs->udp_offloads);
}

/* Callback from net/ipv4/udp.c to receive packets */
static int geneve_udp_encap_recv(struct sock *sk, struct sk_buff *skb)
{
	struct genevehdr *geneveh;
	struct geneve_sock *gs;
	int opts_len;

	/* Need Geneve and inner Ethernet header to be present */
	if (unlikely(!pskb_may_pull(skb, GENEVE_BASE_HLEN)))
		goto error;

	/* Return packets with reserved bits set */
	geneveh = geneve_hdr(skb);

	if (unlikely(geneveh->ver != GENEVE_VER))
		goto error;

	if (unlikely(geneveh->proto_type != htons(ETH_P_TEB)))
		goto error;

	opts_len = geneveh->opt_len * 4;
	if (iptunnel_pull_header(skb, GENEVE_BASE_HLEN + opts_len,
				 htons(ETH_P_TEB)))
		goto drop;

	gs = rcu_dereference_sk_user_data(sk);
	if (!gs)
		goto drop;

	gs->rcv(gs, skb);
	return 0;

drop:
	/* Consume bad packet */
	kfree_skb(skb);
	return 0;

error:
	/* Let the UDP layer deal with the skb */
	return 1;
}

static struct socket *geneve_create_sock(struct net *net, bool ipv6,
					 __be16 port)
{
	struct socket *sock;
	struct udp_port_cfg udp_conf;
	int err;

	memset(&udp_conf, 0, sizeof(udp_conf));

	if (ipv6) {
		udp_conf.family = AF_INET6;
	} else {
		udp_conf.family = AF_INET;
		udp_conf.local_ip.s_addr = htonl(INADDR_ANY);
	}

	udp_conf.local_udp_port = port;

	/* Open UDP socket */
	err = udp_sock_create(net, &udp_conf, &sock);
	if (err < 0)
		return ERR_PTR(err);

	return sock;
}

/* Create new listen socket if needed */
static struct geneve_sock *geneve_socket_create(struct net *net, __be16 port,
						geneve_rcv_t *rcv, void *data,
						bool ipv6)
{
	struct geneve_net *gn = net_generic(net, geneve_net_id);
	struct geneve_sock *gs;
	struct socket *sock;
	struct udp_tunnel_sock_cfg tunnel_cfg;

	gs = kzalloc(sizeof(*gs), GFP_KERNEL);
	if (!gs)
		return ERR_PTR(-ENOMEM);

	sock = geneve_create_sock(net, ipv6, port);
	if (IS_ERR(sock)) {
		kfree(gs);
		return ERR_CAST(sock);
	}

	gs->sock = sock;
	gs->refcnt = 1;
	gs->rcv = rcv;
	gs->rcv_data = data;

	/* Initialize the geneve udp offloads structure */
	gs->udp_offloads.port = port;
	gs->udp_offloads.callbacks.gro_receive  = geneve_gro_receive;
	gs->udp_offloads.callbacks.gro_complete = geneve_gro_complete;
	geneve_notify_add_rx_port(gs);

	/* Mark socket as an encapsulation socket */
	tunnel_cfg.sk_user_data = gs;
	tunnel_cfg.encap_type = 1;
	tunnel_cfg.encap_rcv = geneve_udp_encap_recv;
	tunnel_cfg.encap_destroy = NULL;
	setup_udp_tunnel_sock(net, sock, &tunnel_cfg);

	list_add(&gs->list, &gn->sock_list);

	return gs;
}

struct geneve_sock *geneve_sock_add(struct net *net, __be16 port,
				    geneve_rcv_t *rcv, void *data,
				    bool no_share, bool ipv6)
{
	struct geneve_sock *gs;

	mutex_lock(&geneve_mutex);

	gs = geneve_find_sock(net, ipv6 ? AF_INET6 : AF_INET, port);
	if (gs) {
		if (!no_share && gs->rcv == rcv)
			gs->refcnt++;
		else
			gs = ERR_PTR(-EBUSY);
	} else {
		gs = geneve_socket_create(net, port, rcv, data, ipv6);
	}

	mutex_unlock(&geneve_mutex);

	return gs;
}
EXPORT_SYMBOL_GPL(geneve_sock_add);

void geneve_sock_release(struct geneve_sock *gs)
{
	mutex_lock(&geneve_mutex);

	if (--gs->refcnt)
		goto unlock;

	list_del(&gs->list);
	geneve_notify_del_rx_port(gs);
	udp_tunnel_sock_release(gs->sock);
	kfree_rcu(gs, rcu);

unlock:
	mutex_unlock(&geneve_mutex);
}
EXPORT_SYMBOL_GPL(geneve_sock_release);

static __net_init int geneve_init_net(struct net *net)
{
	struct geneve_net *gn = net_generic(net, geneve_net_id);

	INIT_LIST_HEAD(&gn->sock_list);

	return 0;
}

static struct pernet_operations geneve_net_ops = {
	.init = geneve_init_net,
	.id   = &geneve_net_id,
	.size = sizeof(struct geneve_net),
};

static int __init geneve_init_module(void)
{
	int rc;

	rc = register_pernet_subsys(&geneve_net_ops);
	if (rc)
		return rc;

	pr_info("Geneve driver\n");

	return 0;
}
module_init(geneve_init_module);

static void __exit geneve_cleanup_module(void)
{
	unregister_pernet_subsys(&geneve_net_ops);
}
module_exit(geneve_cleanup_module);

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Jesse Gross <jesse@nicira.com>");
MODULE_DESCRIPTION("Driver for GENEVE encapsulated traffic");
MODULE_ALIAS_RTNL_LINK("geneve");
Commit	Line	Data
0b5e8b8e AZ	1	/*
	2	* Geneve: Generic Network Virtualization Encapsulation
	3	*
	4	* Copyright (c) 2014 Nicira, Inc.
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License
	8	* as published by the Free Software Foundation; either version
	9	* 2 of the License, or (at your option) any later version.
	10	*/
	11
	12	#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
	13
	14	#include <linux/kernel.h>
	15	#include <linux/types.h>
	16	#include <linux/module.h>
	17	#include <linux/errno.h>
	18	#include <linux/slab.h>
	19	#include <linux/skbuff.h>
829a3ada	20	#include <linux/list.h>
0b5e8b8e AZ	21	#include <linux/netdevice.h>
	22	#include <linux/in.h>
	23	#include <linux/ip.h>
	24	#include <linux/udp.h>
	25	#include <linux/igmp.h>
	26	#include <linux/etherdevice.h>
	27	#include <linux/if_ether.h>
	28	#include <linux/if_vlan.h>
0b5e8b8e	29	#include <linux/ethtool.h>
829a3ada	30	#include <linux/mutex.h>
0b5e8b8e AZ	31	#include <net/arp.h>
	32	#include <net/ndisc.h>
	33	#include <net/ip.h>
	34	#include <net/ip_tunnels.h>
	35	#include <net/icmp.h>
	36	#include <net/udp.h>
	37	#include <net/rtnetlink.h>
	38	#include <net/route.h>
	39	#include <net/dsfield.h>
	40	#include <net/inet_ecn.h>
	41	#include <net/net_namespace.h>
	42	#include <net/netns/generic.h>
	43	#include <net/geneve.h>
	44	#include <net/protocol.h>
	45	#include <net/udp_tunnel.h>
	46	#if IS_ENABLED(CONFIG_IPV6)
	47	#include <net/ipv6.h>
	48	#include <net/addrconf.h>
	49	#include <net/ip6_tunnel.h>
	50	#include <net/ip6_checksum.h>
	51	#endif
	52
829a3ada JG	53	/* Protects sock_list and refcounts. */
	54	static DEFINE_MUTEX(geneve_mutex);
	55
0b5e8b8e AZ	56	/* per-network namespace private data for this module */
0b5e8b8e AZ	57	struct geneve_net {
df5dba8e	58	struct list_head sock_list;
0b5e8b8e AZ	59	};
	60
	61	static int geneve_net_id;
	62
0b5e8b8e AZ	63	static inline struct genevehdr geneve_hdr(const struct sk_buff skb)
	64	{
	65	return (struct genevehdr *)(udp_hdr(skb) + 1);
	66	}
	67
46b1e4f9 JG	68	static struct geneve_sock geneve_find_sock(struct net net,
46b1e4f9 JG	69	sa_family_t family, __be16 port)
0b5e8b8e	70	{
df5dba8e	71	struct geneve_net *gn = net_generic(net, geneve_net_id);
0b5e8b8e AZ	72	struct geneve_sock *gs;
0b5e8b8e AZ	73
df5dba8e	74	list_for_each_entry(gs, &gn->sock_list, list) {
46b1e4f9 JG	75	if (inet_sk(gs->sock->sk)->inet_sport == port &&
46b1e4f9 JG	76	inet_sk(gs->sock->sk)->sk.sk_family == family)
0b5e8b8e AZ	77	return gs;
	78	}
	79
	80	return NULL;
	81	}
	82
	83	static void geneve_build_header(struct genevehdr *geneveh,
	84	__be16 tun_flags, u8 vni[3],
	85	u8 options_len, u8 *options)
	86	{
	87	geneveh->ver = GENEVE_VER;
	88	geneveh->opt_len = options_len / 4;
	89	geneveh->oam = !!(tun_flags & TUNNEL_OAM);
	90	geneveh->critical = !!(tun_flags & TUNNEL_CRIT_OPT);
	91	geneveh->rsvd1 = 0;
	92	memcpy(geneveh->vni, vni, 3);
	93	geneveh->proto_type = htons(ETH_P_TEB);
	94	geneveh->rsvd2 = 0;
	95
	96	memcpy(geneveh->options, options, options_len);
	97	}
	98
f4e715c3	99	/* Transmit a fully formatted Geneve frame.
0b5e8b8e AZ	100	*
	101	* When calling this function. The skb->data should point
	102	* to the geneve header which is fully formed.
	103	*
	104	* This function will add other UDP tunnel headers.
	105	*/
	106	int geneve_xmit_skb(struct geneve_sock gs, struct rtable rt,
	107	struct sk_buff *skb, __be32 src, __be32 dst, __u8 tos,
	108	__u8 ttl, __be16 df, __be16 src_port, __be16 dst_port,
	109	__be16 tun_flags, u8 vni[3], u8 opt_len, u8 *opt,
	110	bool xnet)
	111	{
	112	struct genevehdr *gnvh;
	113	int min_headroom;
	114	int err;
	115
	116	skb = udp_tunnel_handle_offloads(skb, !gs->sock->sk->sk_no_check_tx);
997e068e PS	117	if (IS_ERR(skb))
997e068e PS	118	return PTR_ERR(skb);
0b5e8b8e AZ	119
	120	min_headroom = LL_RESERVED_SPACE(rt->dst.dev) + rt->dst.header_len
	121	+ GENEVE_BASE_HLEN + opt_len + sizeof(struct iphdr)
df8a39de	122	+ (skb_vlan_tag_present(skb) ? VLAN_HLEN : 0);
0b5e8b8e AZ	123
0b5e8b8e AZ	124	err = skb_cow_head(skb, min_headroom);
997e068e PS	125	if (unlikely(err)) {
997e068e PS	126	kfree_skb(skb);
0b5e8b8e	127	return err;
997e068e	128	}
0b5e8b8e	129
5968250c JP	130	skb = vlan_hwaccel_push_inside(skb);
	131	if (unlikely(!skb))
	132	return -ENOMEM;
0b5e8b8e AZ	133
	134	gnvh = (struct genevehdr )__skb_push(skb, sizeof(gnvh) + opt_len);
	135	geneve_build_header(gnvh, tun_flags, vni, opt_len, opt);
	136
45cac46e JG	137	skb_set_inner_protocol(skb, htons(ETH_P_TEB));
45cac46e JG	138
0b5e8b8e AZ	139	return udp_tunnel_xmit_skb(gs->sock, rt, skb, src, dst,
	140	tos, ttl, df, src_port, dst_port, xnet);
	141	}
	142	EXPORT_SYMBOL_GPL(geneve_xmit_skb);
	143
a4c9ea5e JS	144	static int geneve_hlen(struct genevehdr *gh)
	145	{
	146	return sizeof(gh) + gh->opt_len 4;
	147	}
	148
	149	static struct sk_buff geneve_gro_receive(struct sk_buff head,
a2b12f3c TH	150	struct sk_buff *skb,
a2b12f3c TH	151	struct udp_offload *uoff)
a4c9ea5e JS	152	{
	153	struct sk_buff p, *pp = NULL;
	154	struct genevehdr gh, gh2;
	155	unsigned int hlen, gh_len, off_gnv;
	156	const struct packet_offload *ptype;
	157	__be16 type;
	158	int flush = 1;
	159
	160	off_gnv = skb_gro_offset(skb);
	161	hlen = off_gnv + sizeof(*gh);
	162	gh = skb_gro_header_fast(skb, off_gnv);
	163	if (skb_gro_header_hard(skb, hlen)) {
	164	gh = skb_gro_header_slow(skb, hlen, off_gnv);
	165	if (unlikely(!gh))
	166	goto out;
	167	}
	168
	169	if (gh->ver != GENEVE_VER \|\| gh->oam)
	170	goto out;
	171	gh_len = geneve_hlen(gh);
	172
	173	hlen = off_gnv + gh_len;
	174	if (skb_gro_header_hard(skb, hlen)) {
	175	gh = skb_gro_header_slow(skb, hlen, off_gnv);
	176	if (unlikely(!gh))
	177	goto out;
	178	}
	179
	180	flush = 0;
	181
	182	for (p = *head; p; p = p->next) {
	183	if (!NAPI_GRO_CB(p)->same_flow)
	184	continue;
	185
	186	gh2 = (struct genevehdr *)(p->data + off_gnv);
	187	if (gh->opt_len != gh2->opt_len \|\|
	188	memcmp(gh, gh2, gh_len)) {
	189	NAPI_GRO_CB(p)->same_flow = 0;
	190	continue;
	191	}
	192	}
	193
	194	type = gh->proto_type;
	195
	196	rcu_read_lock();
	197	ptype = gro_find_receive_by_type(type);
	198	if (ptype == NULL) {
	199	flush = 1;
	200	goto out_unlock;
	201	}
	202
	203	skb_gro_pull(skb, gh_len);
	204	skb_gro_postpull_rcsum(skb, gh, gh_len);
	205	pp = ptype->callbacks.gro_receive(head, skb);
	206
	207	out_unlock:
	208	rcu_read_unlock();
	209	out:
	210	NAPI_GRO_CB(skb)->flush \|= flush;
	211
	212	return pp;
	213	}
	214
a2b12f3c TH	215	static int geneve_gro_complete(struct sk_buff *skb, int nhoff,
a2b12f3c TH	216	struct udp_offload *uoff)
a4c9ea5e JS	217	{
	218	struct genevehdr *gh;
	219	struct packet_offload *ptype;
	220	__be16 type;
	221	int gh_len;
	222	int err = -ENOSYS;
	223
	224	udp_tunnel_gro_complete(skb, nhoff);
	225
	226	gh = (struct genevehdr *)(skb->data + nhoff);
	227	gh_len = geneve_hlen(gh);
	228	type = gh->proto_type;
	229
	230	rcu_read_lock();
	231	ptype = gro_find_complete_by_type(type);
	232	if (ptype != NULL)
	233	err = ptype->callbacks.gro_complete(skb, nhoff + gh_len);
	234
	235	rcu_read_unlock();
	236	return err;
	237	}
	238
0b5e8b8e AZ	239	static void geneve_notify_add_rx_port(struct geneve_sock *gs)
	240	{
	241	struct sock *sk = gs->sock->sk;
	242	sa_family_t sa_family = sk->sk_family;
	243	int err;
	244
	245	if (sa_family == AF_INET) {
	246	err = udp_add_offload(&gs->udp_offloads);
	247	if (err)
	248	pr_warn("geneve: udp_add_offload failed with status %d\n",
	249	err);
	250	}
	251	}
	252
7ed767f7 JG	253	static void geneve_notify_del_rx_port(struct geneve_sock *gs)
	254	{
	255	struct sock *sk = gs->sock->sk;
	256	sa_family_t sa_family = sk->sk_family;
	257
	258	if (sa_family == AF_INET)
	259	udp_del_offload(&gs->udp_offloads);
	260	}
	261
0b5e8b8e AZ	262	/* Callback from net/ipv4/udp.c to receive packets */
	263	static int geneve_udp_encap_recv(struct sock sk, struct sk_buff skb)
	264	{
	265	struct genevehdr *geneveh;
	266	struct geneve_sock *gs;
	267	int opts_len;
	268
	269	/* Need Geneve and inner Ethernet header to be present */
	270	if (unlikely(!pskb_may_pull(skb, GENEVE_BASE_HLEN)))
	271	goto error;
	272
	273	/* Return packets with reserved bits set */
	274	geneveh = geneve_hdr(skb);
	275
	276	if (unlikely(geneveh->ver != GENEVE_VER))
	277	goto error;
	278
	279	if (unlikely(geneveh->proto_type != htons(ETH_P_TEB)))
	280	goto error;
	281
	282	opts_len = geneveh->opt_len * 4;
	283	if (iptunnel_pull_header(skb, GENEVE_BASE_HLEN + opts_len,
	284	htons(ETH_P_TEB)))
	285	goto drop;
	286
	287	gs = rcu_dereference_sk_user_data(sk);
	288	if (!gs)
	289	goto drop;
	290
	291	gs->rcv(gs, skb);
	292	return 0;
	293
	294	drop:
	295	/* Consume bad packet */
	296	kfree_skb(skb);
	297	return 0;
	298
	299	error:
	300	/* Let the UDP layer deal with the skb */
	301	return 1;
	302	}
	303
0b5e8b8e AZ	304	static struct socket geneve_create_sock(struct net net, bool ipv6,
	305	__be16 port)
	306	{
	307	struct socket *sock;
	308	struct udp_port_cfg udp_conf;
	309	int err;
	310
	311	memset(&udp_conf, 0, sizeof(udp_conf));
	312
	313	if (ipv6) {
	314	udp_conf.family = AF_INET6;
	315	} else {
	316	udp_conf.family = AF_INET;
42350dca	317	udp_conf.local_ip.s_addr = htonl(INADDR_ANY);
0b5e8b8e AZ	318	}
	319
	320	udp_conf.local_udp_port = port;
	321
	322	/* Open UDP socket */
	323	err = udp_sock_create(net, &udp_conf, &sock);
	324	if (err < 0)
	325	return ERR_PTR(err);
	326
	327	return sock;
	328	}
	329
	330	/* Create new listen socket if needed */
	331	static struct geneve_sock geneve_socket_create(struct net net, __be16 port,
	332	geneve_rcv_t rcv, void data,
	333	bool ipv6)
	334	{
df5dba8e	335	struct geneve_net *gn = net_generic(net, geneve_net_id);
0b5e8b8e AZ	336	struct geneve_sock *gs;
	337	struct socket *sock;
	338	struct udp_tunnel_sock_cfg tunnel_cfg;
	339
	340	gs = kzalloc(sizeof(*gs), GFP_KERNEL);
	341	if (!gs)
	342	return ERR_PTR(-ENOMEM);
	343
0b5e8b8e AZ	344	sock = geneve_create_sock(net, ipv6, port);
	345	if (IS_ERR(sock)) {
	346	kfree(gs);
	347	return ERR_CAST(sock);
	348	}
	349
	350	gs->sock = sock;
829a3ada	351	gs->refcnt = 1;
0b5e8b8e AZ	352	gs->rcv = rcv;
	353	gs->rcv_data = data;
	354
	355	/* Initialize the geneve udp offloads structure */
	356	gs->udp_offloads.port = port;
a4c9ea5e JS	357	gs->udp_offloads.callbacks.gro_receive = geneve_gro_receive;
a4c9ea5e JS	358	gs->udp_offloads.callbacks.gro_complete = geneve_gro_complete;
0b5e8b8e	359	geneve_notify_add_rx_port(gs);
0b5e8b8e AZ	360
	361	/* Mark socket as an encapsulation socket */
	362	tunnel_cfg.sk_user_data = gs;
	363	tunnel_cfg.encap_type = 1;
	364	tunnel_cfg.encap_rcv = geneve_udp_encap_recv;
	365	tunnel_cfg.encap_destroy = NULL;
	366	setup_udp_tunnel_sock(net, sock, &tunnel_cfg);
	367
df5dba8e	368	list_add(&gs->list, &gn->sock_list);
829a3ada	369
0b5e8b8e AZ	370	return gs;
	371	}
	372
	373	struct geneve_sock geneve_sock_add(struct net net, __be16 port,
	374	geneve_rcv_t rcv, void data,
	375	bool no_share, bool ipv6)
	376	{
	377	struct geneve_sock *gs;
	378
829a3ada	379	mutex_lock(&geneve_mutex);
0b5e8b8e	380
46b1e4f9	381	gs = geneve_find_sock(net, ipv6 ? AF_INET6 : AF_INET, port);
829a3ada JG	382	if (gs) {
	383	if (!no_share && gs->rcv == rcv)
	384	gs->refcnt++;
	385	else
0b5e8b8e	386	gs = ERR_PTR(-EBUSY);
829a3ada JG	387	} else {
	388	gs = geneve_socket_create(net, port, rcv, data, ipv6);
	389	}
12069401	390
829a3ada	391	mutex_unlock(&geneve_mutex);
0b5e8b8e AZ	392
	393	return gs;
	394	}
	395	EXPORT_SYMBOL_GPL(geneve_sock_add);
	396
	397	void geneve_sock_release(struct geneve_sock *gs)
	398	{
829a3ada	399	mutex_lock(&geneve_mutex);
7ed767f7	400
829a3ada JG	401	if (--gs->refcnt)
829a3ada JG	402	goto unlock;
0b5e8b8e	403
df5dba8e	404	list_del(&gs->list);
7ed767f7	405	geneve_notify_del_rx_port(gs);
61f3cade JG	406	udp_tunnel_sock_release(gs->sock);
61f3cade JG	407	kfree_rcu(gs, rcu);
829a3ada JG	408
	409	unlock:
	410	mutex_unlock(&geneve_mutex);
0b5e8b8e AZ	411	}
	412	EXPORT_SYMBOL_GPL(geneve_sock_release);
	413
	414	static __net_init int geneve_init_net(struct net *net)
	415	{
	416	struct geneve_net *gn = net_generic(net, geneve_net_id);
0b5e8b8e	417
df5dba8e	418	INIT_LIST_HEAD(&gn->sock_list);
0b5e8b8e AZ	419
	420	return 0;
	421	}
	422
	423	static struct pernet_operations geneve_net_ops = {
	424	.init = geneve_init_net,
0b5e8b8e AZ	425	.id = &geneve_net_id,
	426	.size = sizeof(struct geneve_net),
	427	};
	428
	429	static int __init geneve_init_module(void)
	430	{
	431	int rc;
	432
0b5e8b8e AZ	433	rc = register_pernet_subsys(&geneve_net_ops);
	434	if (rc)
	435	return rc;
	436
	437	pr_info("Geneve driver\n");
	438
	439	return 0;
	440	}
829a3ada	441	module_init(geneve_init_module);
0b5e8b8e AZ	442
	443	static void __exit geneve_cleanup_module(void)
	444	{
d3ca9eaf	445	unregister_pernet_subsys(&geneve_net_ops);
0b5e8b8e AZ	446	}
	447	module_exit(geneve_cleanup_module);
	448
	449	MODULE_LICENSE("GPL");
	450	MODULE_AUTHOR("Jesse Gross <jesse@nicira.com>");
	451	MODULE_DESCRIPTION("Driver for GENEVE encapsulated traffic");
	452	MODULE_ALIAS_RTNL_LINK("geneve");