[deliverable/linux.git] / include / net / ip_tunnels.h

#ifndef __NET_IP_TUNNELS_H
#define __NET_IP_TUNNELS_H 1

#include <linux/if_tunnel.h>
#include <linux/netdevice.h>
#include <linux/skbuff.h>
#include <linux/socket.h>
#include <linux/types.h>
#include <linux/u64_stats_sync.h>
#include <linux/bitops.h>

#include <net/dsfield.h>
#include <net/gro_cells.h>
#include <net/inet_ecn.h>
#include <net/netns/generic.h>
#include <net/rtnetlink.h>
#include <net/lwtunnel.h>
#include <net/dst_cache.h>

#if IS_ENABLED(CONFIG_IPV6)
#include <net/ipv6.h>
#include <net/ip6_fib.h>
#include <net/ip6_route.h>
#endif

/* Keep error state on tunnel for 30 sec */
#define IPTUNNEL_ERR_TIMEO	(30*HZ)

/* Used to memset ip_tunnel padding. */
#define IP_TUNNEL_KEY_SIZE	offsetofend(struct ip_tunnel_key, tp_dst)

/* Used to memset ipv4 address padding. */
#define IP_TUNNEL_KEY_IPV4_PAD	offsetofend(struct ip_tunnel_key, u.ipv4.dst)
#define IP_TUNNEL_KEY_IPV4_PAD_LEN				\
	(FIELD_SIZEOF(struct ip_tunnel_key, u) -		\
	 FIELD_SIZEOF(struct ip_tunnel_key, u.ipv4))

struct ip_tunnel_key {
	__be64			tun_id;
	union {
		struct {
			__be32	src;
			__be32	dst;
		} ipv4;
		struct {
			struct in6_addr src;
			struct in6_addr dst;
		} ipv6;
	} u;
	__be16			tun_flags;
	u8			tos;		/* TOS for IPv4, TC for IPv6 */
	u8			ttl;		/* TTL for IPv4, HL for IPv6 */
	__be32			label;		/* Flow Label for IPv6 */
	__be16			tp_src;
	__be16			tp_dst;
};

/* Flags for ip_tunnel_info mode. */
#define IP_TUNNEL_INFO_TX	0x01	/* represents tx tunnel parameters */
#define IP_TUNNEL_INFO_IPV6	0x02	/* key contains IPv6 addresses */

/* Maximum tunnel options length. */
#define IP_TUNNEL_OPTS_MAX					\
	GENMASK((FIELD_SIZEOF(struct ip_tunnel_info,		\
			      options_len) * BITS_PER_BYTE) - 1, 0)

struct ip_tunnel_info {
	struct ip_tunnel_key	key;
#ifdef CONFIG_DST_CACHE
	struct dst_cache	dst_cache;
#endif
	u8			options_len;
	u8			mode;
};

/* 6rd prefix/relay information */
#ifdef CONFIG_IPV6_SIT_6RD
struct ip_tunnel_6rd_parm {
	struct in6_addr		prefix;
	__be32			relay_prefix;
	u16			prefixlen;
	u16			relay_prefixlen;
};
#endif

struct ip_tunnel_encap {
	u16			type;
	u16			flags;
	__be16			sport;
	__be16			dport;
};

struct ip_tunnel_prl_entry {
	struct ip_tunnel_prl_entry __rcu *next;
	__be32				addr;
	u16				flags;
	struct rcu_head			rcu_head;
};

struct metadata_dst;

struct ip_tunnel {
	struct ip_tunnel __rcu	*next;
	struct hlist_node hash_node;
	struct net_device	*dev;
	struct net		*net;	/* netns for packet i/o */

	unsigned long	err_time;	/* Time when the last ICMP error
					 * arrived */
	int		err_count;	/* Number of arrived ICMP errors */

	/* These four fields used only by GRE */
	u32		i_seqno;	/* The last seen seqno	*/
	u32		o_seqno;	/* The last output seqno */
	int		tun_hlen;	/* Precalculated header length */

	struct dst_cache dst_cache;

	struct ip_tunnel_parm parms;

	int		mlink;
	int		encap_hlen;	/* Encap header length (FOU,GUE) */
	int		hlen;		/* tun_hlen + encap_hlen */
	struct ip_tunnel_encap encap;

	/* for SIT */
#ifdef CONFIG_IPV6_SIT_6RD
	struct ip_tunnel_6rd_parm ip6rd;
#endif
	struct ip_tunnel_prl_entry __rcu *prl;	/* potential router list */
	unsigned int		prl_count;	/* # of entries in PRL */
	int			ip_tnl_net_id;
	struct gro_cells	gro_cells;
	bool			collect_md;
	bool			ignore_df;
};

#define TUNNEL_CSUM		__cpu_to_be16(0x01)
#define TUNNEL_ROUTING		__cpu_to_be16(0x02)
#define TUNNEL_KEY		__cpu_to_be16(0x04)
#define TUNNEL_SEQ		__cpu_to_be16(0x08)
#define TUNNEL_STRICT		__cpu_to_be16(0x10)
#define TUNNEL_REC		__cpu_to_be16(0x20)
#define TUNNEL_VERSION		__cpu_to_be16(0x40)
#define TUNNEL_NO_KEY		__cpu_to_be16(0x80)
#define TUNNEL_DONT_FRAGMENT    __cpu_to_be16(0x0100)
#define TUNNEL_OAM		__cpu_to_be16(0x0200)
#define TUNNEL_CRIT_OPT		__cpu_to_be16(0x0400)
#define TUNNEL_GENEVE_OPT	__cpu_to_be16(0x0800)
#define TUNNEL_VXLAN_OPT	__cpu_to_be16(0x1000)
#define TUNNEL_NOCACHE		__cpu_to_be16(0x2000)

#define TUNNEL_OPTIONS_PRESENT	(TUNNEL_GENEVE_OPT | TUNNEL_VXLAN_OPT)

struct tnl_ptk_info {
	__be16 flags;
	__be16 proto;
	__be32 key;
	__be32 seq;
	int hdr_len;
};

#define PACKET_RCVD	0
#define PACKET_REJECT	1
#define PACKET_NEXT	2

#define IP_TNL_HASH_BITS   7
#define IP_TNL_HASH_SIZE   (1 << IP_TNL_HASH_BITS)

struct ip_tunnel_net {
	struct net_device *fb_tunnel_dev;
	struct hlist_head tunnels[IP_TNL_HASH_SIZE];
	struct ip_tunnel __rcu *collect_md_tun;
};

static inline void ip_tunnel_key_init(struct ip_tunnel_key *key,
				      __be32 saddr, __be32 daddr,
				      u8 tos, u8 ttl, __be32 label,
				      __be16 tp_src, __be16 tp_dst,
				      __be64 tun_id, __be16 tun_flags)
{
	key->tun_id = tun_id;
	key->u.ipv4.src = saddr;
	key->u.ipv4.dst = daddr;
	memset((unsigned char *)key + IP_TUNNEL_KEY_IPV4_PAD,
	       0, IP_TUNNEL_KEY_IPV4_PAD_LEN);
	key->tos = tos;
	key->ttl = ttl;
	key->label = label;
	key->tun_flags = tun_flags;

	/* For the tunnel types on the top of IPsec, the tp_src and tp_dst of
	 * the upper tunnel are used.
	 * E.g: GRE over IPSEC, the tp_src and tp_port are zero.
	 */
	key->tp_src = tp_src;
	key->tp_dst = tp_dst;

	/* Clear struct padding. */
	if (sizeof(*key) != IP_TUNNEL_KEY_SIZE)
		memset((unsigned char *)key + IP_TUNNEL_KEY_SIZE,
		       0, sizeof(*key) - IP_TUNNEL_KEY_SIZE);
}

static inline bool
ip_tunnel_dst_cache_usable(const struct sk_buff *skb,
			   const struct ip_tunnel_info *info)
{
	if (skb->mark)
		return false;
	if (!info)
		return true;
	if (info->key.tun_flags & TUNNEL_NOCACHE)
		return false;

	return true;
}

static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info
					       *tun_info)
{
	return tun_info->mode & IP_TUNNEL_INFO_IPV6 ? AF_INET6 : AF_INET;
}

static inline __be64 key32_to_tunnel_id(__be32 key)
{
#ifdef __BIG_ENDIAN
	return (__force __be64)key;
#else
	return (__force __be64)((__force u64)key << 32);
#endif
}

/* Returns the least-significant 32 bits of a __be64. */
static inline __be32 tunnel_id_to_key32(__be64 tun_id)
{
#ifdef __BIG_ENDIAN
	return (__force __be32)tun_id;
#else
	return (__force __be32)((__force u64)tun_id >> 32);
#endif
}

#ifdef CONFIG_INET

int ip_tunnel_init(struct net_device *dev);
void ip_tunnel_uninit(struct net_device *dev);
void  ip_tunnel_dellink(struct net_device *dev, struct list_head *head);
struct net *ip_tunnel_get_link_net(const struct net_device *dev);
int ip_tunnel_get_iflink(const struct net_device *dev);
int ip_tunnel_init_net(struct net *net, int ip_tnl_net_id,
		       struct rtnl_link_ops *ops, char *devname);

void ip_tunnel_delete_net(struct ip_tunnel_net *itn, struct rtnl_link_ops *ops);

void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
		    const struct iphdr *tnl_params, const u8 protocol);
int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd);
int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict);
int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu);

struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev,
						struct rtnl_link_stats64 *tot);
struct ip_tunnel *ip_tunnel_lookup(struct ip_tunnel_net *itn,
				   int link, __be16 flags,
				   __be32 remote, __be32 local,
				   __be32 key);

int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb,
		  const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst,
		  bool log_ecn_error);
int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[],
			 struct ip_tunnel_parm *p);
int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[],
		      struct ip_tunnel_parm *p);
void ip_tunnel_setup(struct net_device *dev, int net_id);

struct ip_tunnel_encap_ops {
	size_t (*encap_hlen)(struct ip_tunnel_encap *e);
	int (*build_header)(struct sk_buff *skb, struct ip_tunnel_encap *e,
			    u8 *protocol, struct flowi4 *fl4);
};

#define MAX_IPTUN_ENCAP_OPS 8

extern const struct ip_tunnel_encap_ops __rcu *
		iptun_encaps[MAX_IPTUN_ENCAP_OPS];

int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op,
			    unsigned int num);
int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
			    unsigned int num);

int ip_tunnel_encap_setup(struct ip_tunnel *t,
			  struct ip_tunnel_encap *ipencap);

static inline int ip_encap_hlen(struct ip_tunnel_encap *e)
{
	const struct ip_tunnel_encap_ops *ops;
	int hlen = -EINVAL;

	if (e->type == TUNNEL_ENCAP_NONE)
		return 0;

	if (e->type >= MAX_IPTUN_ENCAP_OPS)
		return -EINVAL;

	rcu_read_lock();
	ops = rcu_dereference(iptun_encaps[e->type]);
	if (likely(ops && ops->encap_hlen))
		hlen = ops->encap_hlen(e);
	rcu_read_unlock();

	return hlen;
}

static inline int ip_tunnel_encap(struct sk_buff *skb, struct ip_tunnel *t,
				  u8 *protocol, struct flowi4 *fl4)
{
	const struct ip_tunnel_encap_ops *ops;
	int ret = -EINVAL;

	if (t->encap.type == TUNNEL_ENCAP_NONE)
		return 0;

	if (t->encap.type >= MAX_IPTUN_ENCAP_OPS)
		return -EINVAL;

	rcu_read_lock();
	ops = rcu_dereference(iptun_encaps[t->encap.type]);
	if (likely(ops && ops->build_header))
		ret = ops->build_header(skb, &t->encap, protocol, fl4);
	rcu_read_unlock();

	return ret;
}

/* Extract dsfield from inner protocol */
static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph,
				       const struct sk_buff *skb)
{
	if (skb->protocol == htons(ETH_P_IP))
		return iph->tos;
	else if (skb->protocol == htons(ETH_P_IPV6))
		return ipv6_get_dsfield((const struct ipv6hdr *)iph);
	else
		return 0;
}

/* Propogate ECN bits out */
static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
				     const struct sk_buff *skb)
{
	u8 inner = ip_tunnel_get_dsfield(iph, skb);

	return INET_ECN_encapsulate(tos, inner);
}

int __iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
			   __be16 inner_proto, bool raw_proto, bool xnet);

static inline int iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
				       __be16 inner_proto, bool xnet)
{
	return __iptunnel_pull_header(skb, hdr_len, inner_proto, false, xnet);
}

void iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
		   __be32 src, __be32 dst, u8 proto,
		   u8 tos, u8 ttl, __be16 df, bool xnet);
struct metadata_dst *iptunnel_metadata_reply(struct metadata_dst *md,
					     gfp_t flags);

int iptunnel_handle_offloads(struct sk_buff *skb, int gso_type_mask);

static inline int iptunnel_pull_offloads(struct sk_buff *skb)
{
	if (skb_is_gso(skb)) {
		int err;

		err = skb_unclone(skb, GFP_ATOMIC);
		if (unlikely(err))
			return err;
		skb_shinfo(skb)->gso_type &= ~(NETIF_F_GSO_ENCAP_ALL >>
					       NETIF_F_GSO_SHIFT);
	}

	skb->encapsulation = 0;
	return 0;
}

static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len)
{
	if (pkt_len > 0) {
		struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats);

		u64_stats_update_begin(&tstats->syncp);
		tstats->tx_bytes += pkt_len;
		tstats->tx_packets++;
		u64_stats_update_end(&tstats->syncp);
		put_cpu_ptr(tstats);
	} else {
		struct net_device_stats *err_stats = &dev->stats;

		if (pkt_len < 0) {
			err_stats->tx_errors++;
			err_stats->tx_aborted_errors++;
		} else {
			err_stats->tx_dropped++;
		}
	}
}

static inline void *ip_tunnel_info_opts(struct ip_tunnel_info *info)
{
	return info + 1;
}

static inline void ip_tunnel_info_opts_get(void *to,
					   const struct ip_tunnel_info *info)
{
	memcpy(to, info + 1, info->options_len);
}

static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
					   const void *from, int len)
{
	memcpy(ip_tunnel_info_opts(info), from, len);
	info->options_len = len;
}

static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
{
	return (struct ip_tunnel_info *)lwtstate->data;
}

extern struct static_key ip_tunnel_metadata_cnt;

/* Returns > 0 if metadata should be collected */
static inline int ip_tunnel_collect_metadata(void)
{
	return static_key_false(&ip_tunnel_metadata_cnt);
}

void __init ip_tunnel_core_init(void);

void ip_tunnel_need_metadata(void);
void ip_tunnel_unneed_metadata(void);

#else /* CONFIG_INET */

static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
{
	return NULL;
}

static inline void ip_tunnel_need_metadata(void)
{
}

static inline void ip_tunnel_unneed_metadata(void)
{
}

static inline void ip_tunnel_info_opts_get(void *to,
					   const struct ip_tunnel_info *info)
{
}

static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
					   const void *from, int len)
{
	info->options_len = 0;
}

#endif /* CONFIG_INET */

#endif /* __NET_IP_TUNNELS_H */
Commit	Line	Data
c5441932 PS	1	#ifndef __NET_IP_TUNNELS_H
	2	#define __NET_IP_TUNNELS_H 1
	3
	4	#include <linux/if_tunnel.h>
	5	#include <linux/netdevice.h>
	6	#include <linux/skbuff.h>
7f9562a1	7	#include <linux/socket.h>
c5441932 PS	8	#include <linux/types.h>
c5441932 PS	9	#include <linux/u64_stats_sync.h>
fca5fdf6 DB	10	#include <linux/bitops.h>
fca5fdf6 DB	11
c5441932 PS	12	#include <net/dsfield.h>
	13	#include <net/gro_cells.h>
	14	#include <net/inet_ecn.h>
56328486	15	#include <net/netns/generic.h>
c5441932	16	#include <net/rtnetlink.h>
3093fbe7	17	#include <net/lwtunnel.h>
e09acddf	18	#include <net/dst_cache.h>
c5441932 PS	19
	20	#if IS_ENABLED(CONFIG_IPV6)
	21	#include <net/ipv6.h>
	22	#include <net/ip6_fib.h>
	23	#include <net/ip6_route.h>
	24	#endif
	25
	26	/* Keep error state on tunnel for 30 sec */
	27	#define IPTUNNEL_ERR_TIMEO (30*HZ)
	28
1d8fff90	29	/* Used to memset ip_tunnel padding. */
376534a3	30	#define IP_TUNNEL_KEY_SIZE offsetofend(struct ip_tunnel_key, tp_dst)
1d8fff90	31
c1ea5d67 JB	32	/* Used to memset ipv4 address padding. */
	33	#define IP_TUNNEL_KEY_IPV4_PAD offsetofend(struct ip_tunnel_key, u.ipv4.dst)
	34	#define IP_TUNNEL_KEY_IPV4_PAD_LEN \
	35	(FIELD_SIZEOF(struct ip_tunnel_key, u) - \
	36	FIELD_SIZEOF(struct ip_tunnel_key, u.ipv4))
	37
1d8fff90 TG	38	struct ip_tunnel_key {
1d8fff90 TG	39	__be64 tun_id;
c1ea5d67 JB	40	union {
	41	struct {
	42	__be32 src;
	43	__be32 dst;
	44	} ipv4;
	45	struct {
	46	struct in6_addr src;
	47	struct in6_addr dst;
	48	} ipv6;
	49	} u;
1d8fff90	50	__be16 tun_flags;
7c383fb2 JB	51	u8 tos; /* TOS for IPv4, TC for IPv6 */
7c383fb2 JB	52	u8 ttl; /* TTL for IPv4, HL for IPv6 */
13461144	53	__be32 label; /* Flow Label for IPv6 */
1d8fff90 TG	54	__be16 tp_src;
1d8fff90 TG	55	__be16 tp_dst;
ac1cf399	56	};
1d8fff90	57
46fa062a JB	58	/* Flags for ip_tunnel_info mode. */
46fa062a JB	59	#define IP_TUNNEL_INFO_TX 0x01 /* represents tx tunnel parameters */
7f9562a1	60	#define IP_TUNNEL_INFO_IPV6 0x02 /* key contains IPv6 addresses */
ee122c79	61
fca5fdf6 DB	62	/* Maximum tunnel options length. */
	63	#define IP_TUNNEL_OPTS_MAX \
	64	GENMASK((FIELD_SIZEOF(struct ip_tunnel_info, \
	65	options_len) * BITS_PER_BYTE) - 1, 0)
	66
1d8fff90 TG	67	struct ip_tunnel_info {
1d8fff90 TG	68	struct ip_tunnel_key key;
d71785ff PA	69	#ifdef CONFIG_DST_CACHE
	70	struct dst_cache dst_cache;
	71	#endif
1d8fff90	72	u8 options_len;
ee122c79	73	u8 mode;
1d8fff90 TG	74	};
1d8fff90 TG	75
c5441932 PS	76	/* 6rd prefix/relay information */
	77	#ifdef CONFIG_IPV6_SIT_6RD
	78	struct ip_tunnel_6rd_parm {
	79	struct in6_addr prefix;
	80	__be32 relay_prefix;
	81	u16 prefixlen;
	82	u16 relay_prefixlen;
	83	};
	84	#endif
	85
56328486	86	struct ip_tunnel_encap {
6b8847c5 JB	87	u16 type;
6b8847c5 JB	88	u16 flags;
56328486 TH	89	__be16 sport;
	90	__be16 dport;
	91	};
	92
c5441932 PS	93	struct ip_tunnel_prl_entry {
	94	struct ip_tunnel_prl_entry __rcu *next;
	95	__be32 addr;
	96	u16 flags;
	97	struct rcu_head rcu_head;
	98	};
	99
2e15ea39 PS	100	struct metadata_dst;
2e15ea39 PS	101
c5441932 PS	102	struct ip_tunnel {
	103	struct ip_tunnel __rcu *next;
	104	struct hlist_node hash_node;
	105	struct net_device *dev;
5e6700b3	106	struct net net; / netns for packet i/o */
c5441932	107
c5441932 PS	108	unsigned long err_time; /* Time when the last ICMP error
c5441932 PS	109	* arrived */
f38ba953	110	int err_count; /* Number of arrived ICMP errors */
c5441932 PS	111
c5441932 PS	112	/* These four fields used only by GRE */
6b8847c5 JB	113	u32 i_seqno; /* The last seen seqno */
6b8847c5 JB	114	u32 o_seqno; /* The last output seqno */
56328486	115	int tun_hlen; /* Precalculated header length */
c5441932	116
e09acddf	117	struct dst_cache dst_cache;
7d442fab	118
c5441932 PS	119	struct ip_tunnel_parm parms;
c5441932 PS	120
f38ba953	121	int mlink;
56328486	122	int encap_hlen; /* Encap header length (FOU,GUE) */
56328486	123	int hlen; /* tun_hlen + encap_hlen */
f38ba953	124	struct ip_tunnel_encap encap;
56328486	125
c5441932 PS	126	/* for SIT */
	127	#ifdef CONFIG_IPV6_SIT_6RD
	128	struct ip_tunnel_6rd_parm ip6rd;
	129	#endif
	130	struct ip_tunnel_prl_entry __rcu prl; / potential router list */
	131	unsigned int prl_count; /* # of entries in PRL */
	132	int ip_tnl_net_id;
	133	struct gro_cells gro_cells;
2e15ea39	134	bool collect_md;
22a59be8	135	bool ignore_df;
c5441932 PS	136	};
c5441932 PS	137
f5796684 JG	138	#define TUNNEL_CSUM __cpu_to_be16(0x01)
	139	#define TUNNEL_ROUTING __cpu_to_be16(0x02)
	140	#define TUNNEL_KEY __cpu_to_be16(0x04)
	141	#define TUNNEL_SEQ __cpu_to_be16(0x08)
	142	#define TUNNEL_STRICT __cpu_to_be16(0x10)
	143	#define TUNNEL_REC __cpu_to_be16(0x20)
	144	#define TUNNEL_VERSION __cpu_to_be16(0x40)
	145	#define TUNNEL_NO_KEY __cpu_to_be16(0x80)
9a628224	146	#define TUNNEL_DONT_FRAGMENT __cpu_to_be16(0x0100)
f5796684 JG	147	#define TUNNEL_OAM __cpu_to_be16(0x0200)
f5796684 JG	148	#define TUNNEL_CRIT_OPT __cpu_to_be16(0x0400)
1dd144cf TG	149	#define TUNNEL_GENEVE_OPT __cpu_to_be16(0x0800)
1dd144cf TG	150	#define TUNNEL_VXLAN_OPT __cpu_to_be16(0x1000)
db3c6139	151	#define TUNNEL_NOCACHE __cpu_to_be16(0x2000)
1dd144cf TG	152
1dd144cf TG	153	#define TUNNEL_OPTIONS_PRESENT (TUNNEL_GENEVE_OPT \| TUNNEL_VXLAN_OPT)
c5441932 PS	154
	155	struct tnl_ptk_info {
	156	__be16 flags;
	157	__be16 proto;
	158	__be32 key;
	159	__be32 seq;
9b8c6d7b	160	int hdr_len;
c5441932 PS	161	};
	162
	163	#define PACKET_RCVD 0
	164	#define PACKET_REJECT 1
125372fa	165	#define PACKET_NEXT 2
c5441932	166
6261d983	167	#define IP_TNL_HASH_BITS 7
c5441932 PS	168	#define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS)
	169
	170	struct ip_tunnel_net {
c5441932	171	struct net_device *fb_tunnel_dev;
6261d983	172	struct hlist_head tunnels[IP_TNL_HASH_SIZE];
2e15ea39	173	struct ip_tunnel __rcu *collect_md_tun;
c5441932 PS	174	};
c5441932 PS	175
4c222798 PS	176	static inline void ip_tunnel_key_init(struct ip_tunnel_key *key,
4c222798 PS	177	__be32 saddr, __be32 daddr,
13461144	178	u8 tos, u8 ttl, __be32 label,
4c222798 PS	179	__be16 tp_src, __be16 tp_dst,
4c222798 PS	180	__be64 tun_id, __be16 tun_flags)
1d8fff90	181	{
4c222798 PS	182	key->tun_id = tun_id;
	183	key->u.ipv4.src = saddr;
	184	key->u.ipv4.dst = daddr;
	185	memset((unsigned char *)key + IP_TUNNEL_KEY_IPV4_PAD,
c1ea5d67	186	0, IP_TUNNEL_KEY_IPV4_PAD_LEN);
4c222798 PS	187	key->tos = tos;
4c222798 PS	188	key->ttl = ttl;
13461144	189	key->label = label;
4c222798	190	key->tun_flags = tun_flags;
1d8fff90 TG	191
	192	/* For the tunnel types on the top of IPsec, the tp_src and tp_dst of
	193	* the upper tunnel are used.
	194	* E.g: GRE over IPSEC, the tp_src and tp_port are zero.
	195	*/
4c222798 PS	196	key->tp_src = tp_src;
4c222798 PS	197	key->tp_dst = tp_dst;
1d8fff90 TG	198
1d8fff90 TG	199	/* Clear struct padding. */
4c222798 PS	200	if (sizeof(*key) != IP_TUNNEL_KEY_SIZE)
	201	memset((unsigned char *)key + IP_TUNNEL_KEY_SIZE,
	202	0, sizeof(*key) - IP_TUNNEL_KEY_SIZE);
1d8fff90 TG	203	}
1d8fff90 TG	204
db3c6139 DB	205	static inline bool
	206	ip_tunnel_dst_cache_usable(const struct sk_buff *skb,
	207	const struct ip_tunnel_info *info)
	208	{
	209	if (skb->mark)
	210	return false;
	211	if (!info)
	212	return true;
	213	if (info->key.tun_flags & TUNNEL_NOCACHE)
	214	return false;
	215
	216	return true;
	217	}
	218
7f9562a1 JB	219	static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info
	220	*tun_info)
	221	{
	222	return tun_info->mode & IP_TUNNEL_INFO_IPV6 ? AF_INET6 : AF_INET;
	223	}
	224
d817f432 AV	225	static inline __be64 key32_to_tunnel_id(__be32 key)
	226	{
	227	#ifdef __BIG_ENDIAN
	228	return (__force __be64)key;
	229	#else
	230	return (__force __be64)((__force u64)key << 32);
	231	#endif
	232	}
	233
	234	/* Returns the least-significant 32 bits of a __be64. */
	235	static inline __be32 tunnel_id_to_key32(__be64 tun_id)
	236	{
	237	#ifdef __BIG_ENDIAN
	238	return (__force __be32)tun_id;
	239	#else
	240	return (__force __be32)((__force u64)tun_id >> 32);
	241	#endif
	242	}
	243
5243b6ac JG	244	#ifdef CONFIG_INET
5243b6ac JG	245
c5441932 PS	246	int ip_tunnel_init(struct net_device *dev);
	247	void ip_tunnel_uninit(struct net_device *dev);
	248	void ip_tunnel_dellink(struct net_device dev, struct list_head head);
1728d4fa	249	struct net ip_tunnel_get_link_net(const struct net_device dev);
1e99584b	250	int ip_tunnel_get_iflink(const struct net_device *dev);
d3b6f614 ED	251	int ip_tunnel_init_net(struct net *net, int ip_tnl_net_id,
d3b6f614 ED	252	struct rtnl_link_ops ops, char devname);
c5441932	253
6c742e71	254	void ip_tunnel_delete_net(struct ip_tunnel_net itn, struct rtnl_link_ops ops);
c5441932 PS	255
c5441932 PS	256	void ip_tunnel_xmit(struct sk_buff skb, struct net_device dev,
bf3d6a8f	257	const struct iphdr *tnl_params, const u8 protocol);
c5441932	258	int ip_tunnel_ioctl(struct net_device dev, struct ip_tunnel_parm p, int cmd);
7e059158	259	int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict);
c5441932 PS	260	int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu);
	261
	262	struct rtnl_link_stats64 ip_tunnel_get_stats64(struct net_device dev,
	263	struct rtnl_link_stats64 *tot);
	264	struct ip_tunnel ip_tunnel_lookup(struct ip_tunnel_net itn,
	265	int link, __be16 flags,
	266	__be32 remote, __be32 local,
	267	__be32 key);
	268
	269	int ip_tunnel_rcv(struct ip_tunnel tunnel, struct sk_buff skb,
2e15ea39 PS	270	const struct tnl_ptk_info tpi, struct metadata_dst tun_dst,
2e15ea39 PS	271	bool log_ecn_error);
c5441932 PS	272	int ip_tunnel_changelink(struct net_device dev, struct nlattr tb[],
	273	struct ip_tunnel_parm *p);
	274	int ip_tunnel_newlink(struct net_device dev, struct nlattr tb[],
	275	struct ip_tunnel_parm *p);
	276	void ip_tunnel_setup(struct net_device *dev, int net_id);
55c2bc14 TH	277
	278	struct ip_tunnel_encap_ops {
	279	size_t (encap_hlen)(struct ip_tunnel_encap e);
	280	int (build_header)(struct sk_buff skb, struct ip_tunnel_encap *e,
	281	u8 protocol, struct flowi4 fl4);
	282	};
	283
	284	#define MAX_IPTUN_ENCAP_OPS 8
	285
	286	extern const struct ip_tunnel_encap_ops __rcu *
	287	iptun_encaps[MAX_IPTUN_ENCAP_OPS];
	288
	289	int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op,
	290	unsigned int num);
	291	int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
	292	unsigned int num);
	293
56328486 TH	294	int ip_tunnel_encap_setup(struct ip_tunnel *t,
56328486 TH	295	struct ip_tunnel_encap *ipencap);
c5441932	296
55c2bc14 TH	297	static inline int ip_encap_hlen(struct ip_tunnel_encap *e)
	298	{
	299	const struct ip_tunnel_encap_ops *ops;
	300	int hlen = -EINVAL;
	301
	302	if (e->type == TUNNEL_ENCAP_NONE)
	303	return 0;
	304
	305	if (e->type >= MAX_IPTUN_ENCAP_OPS)
	306	return -EINVAL;
	307
	308	rcu_read_lock();
	309	ops = rcu_dereference(iptun_encaps[e->type]);
	310	if (likely(ops && ops->encap_hlen))
	311	hlen = ops->encap_hlen(e);
	312	rcu_read_unlock();
	313
	314	return hlen;
	315	}
	316
	317	static inline int ip_tunnel_encap(struct sk_buff skb, struct ip_tunnel t,
	318	u8 protocol, struct flowi4 fl4)
	319	{
	320	const struct ip_tunnel_encap_ops *ops;
	321	int ret = -EINVAL;
	322
	323	if (t->encap.type == TUNNEL_ENCAP_NONE)
	324	return 0;
	325
	326	if (t->encap.type >= MAX_IPTUN_ENCAP_OPS)
	327	return -EINVAL;
	328
	329	rcu_read_lock();
	330	ops = rcu_dereference(iptun_encaps[t->encap.type]);
	331	if (likely(ops && ops->build_header))
	332	ret = ops->build_header(skb, &t->encap, protocol, fl4);
	333	rcu_read_unlock();
	334
	335	return ret;
	336	}
	337
c5441932 PS	338	/* Extract dsfield from inner protocol */
	339	static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph,
	340	const struct sk_buff *skb)
	341	{
	342	if (skb->protocol == htons(ETH_P_IP))
	343	return iph->tos;
	344	else if (skb->protocol == htons(ETH_P_IPV6))
	345	return ipv6_get_dsfield((const struct ipv6hdr *)iph);
	346	else
	347	return 0;
	348	}
	349
	350	/* Propogate ECN bits out */
	351	static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
	352	const struct sk_buff *skb)
	353	{
	354	u8 inner = ip_tunnel_get_dsfield(iph, skb);
	355
	356	return INET_ECN_encapsulate(tos, inner);
	357	}
	358
a6d5bbf3 JB	359	int __iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
	360	__be16 inner_proto, bool raw_proto, bool xnet);
	361
	362	static inline int iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
	363	__be16 inner_proto, bool xnet)
	364	{
	365	return __iptunnel_pull_header(skb, hdr_len, inner_proto, false, xnet);
	366	}
	367
039f5062 PS	368	void iptunnel_xmit(struct sock sk, struct rtable rt, struct sk_buff *skb,
	369	__be32 src, __be32 dst, u8 proto,
	370	u8 tos, u8 ttl, __be16 df, bool xnet);
63d008a4 JB	371	struct metadata_dst iptunnel_metadata_reply(struct metadata_dst md,
63d008a4 JB	372	gfp_t flags);
0e6fbc5b	373
aed069df	374	int iptunnel_handle_offloads(struct sk_buff *skb, int gso_type_mask);
2d26f0a3	375
a09a4c8d JG	376	static inline int iptunnel_pull_offloads(struct sk_buff *skb)
	377	{
	378	if (skb_is_gso(skb)) {
	379	int err;
	380
	381	err = skb_unclone(skb, GFP_ATOMIC);
	382	if (unlikely(err))
	383	return err;
	384	skb_shinfo(skb)->gso_type &= ~(NETIF_F_GSO_ENCAP_ALL >>
	385	NETIF_F_GSO_SHIFT);
	386	}
	387
	388	skb->encapsulation = 0;
	389	return 0;
	390	}
	391
039f5062	392	static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len)
c5441932	393	{
039f5062 PS	394	if (pkt_len > 0) {
039f5062 PS	395	struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats);
c5441932	396
c5441932	397	u64_stats_update_begin(&tstats->syncp);
039f5062	398	tstats->tx_bytes += pkt_len;
c5441932 PS	399	tstats->tx_packets++;
c5441932 PS	400	u64_stats_update_end(&tstats->syncp);
b4fe85f9	401	put_cpu_ptr(tstats);
c5441932	402	} else {
039f5062 PS	403	struct net_device_stats *err_stats = &dev->stats;
	404
	405	if (pkt_len < 0) {
	406	err_stats->tx_errors++;
	407	err_stats->tx_aborted_errors++;
	408	} else {
	409	err_stats->tx_dropped++;
	410	}
c5441932 PS	411	}
c5441932 PS	412	}
5243b6ac	413
4c222798	414	static inline void ip_tunnel_info_opts(struct ip_tunnel_info info)
ee122c79 TG	415	{
	416	return info + 1;
	417	}
	418
4c222798 PS	419	static inline void ip_tunnel_info_opts_get(void *to,
	420	const struct ip_tunnel_info *info)
	421	{
	422	memcpy(to, info + 1, info->options_len);
	423	}
	424
	425	static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
	426	const void *from, int len)
	427	{
	428	memcpy(ip_tunnel_info_opts(info), from, len);
	429	info->options_len = len;
	430	}
	431
3093fbe7 TG	432	static inline struct ip_tunnel_info lwt_tun_info(struct lwtunnel_state lwtstate)
	433	{
	434	return (struct ip_tunnel_info *)lwtstate->data;
	435	}
	436
e7030878 TG	437	extern struct static_key ip_tunnel_metadata_cnt;
	438
	439	/* Returns > 0 if metadata should be collected */
	440	static inline int ip_tunnel_collect_metadata(void)
	441	{
	442	return static_key_false(&ip_tunnel_metadata_cnt);
	443	}
	444
045a0fa0 TG	445	void __init ip_tunnel_core_init(void);
045a0fa0 TG	446
e7030878 TG	447	void ip_tunnel_need_metadata(void);
	448	void ip_tunnel_unneed_metadata(void);
	449
05283187 TG	450	#else /* CONFIG_INET */
	451
	452	static inline struct ip_tunnel_info lwt_tun_info(struct lwtunnel_state lwtstate)
	453	{
	454	return NULL;
	455	}
	456
	457	static inline void ip_tunnel_need_metadata(void)
	458	{
	459	}
	460
	461	static inline void ip_tunnel_unneed_metadata(void)
	462	{
	463	}
	464
e28e87ed DB	465	static inline void ip_tunnel_info_opts_get(void *to,
	466	const struct ip_tunnel_info *info)
	467	{
	468	}
	469
	470	static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
	471	const void *from, int len)
	472	{
	473	info->options_len = 0;
	474	}
	475
5243b6ac JG	476	#endif /* CONFIG_INET */
5243b6ac JG	477
c5441932	478	#endif /* __NET_IP_TUNNELS_H */