[mirror_ubuntu-eoan-kernel.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 */
#define IP_TUNNEL_INFO_BRIDGE	0x04	/* represents a bridged tunnel id */

/* 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 */

	/* This field used only by ERSPAN */
	u32		index;		/* ERSPAN type II index */

	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 */
	unsigned int		ip_tnl_net_id;
	struct gro_cells	gro_cells;
	__u32			fwmark;
	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_ERSPAN_OPT	__cpu_to_be16(0x4000)

#define TUNNEL_OPTIONS_PRESENT \
		(TUNNEL_GENEVE_OPT | TUNNEL_VXLAN_OPT | TUNNEL_ERSPAN_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, unsigned 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);
void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
		       const u8 proto);
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);

void 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, __u32 fwmark);
int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[],
		      struct ip_tunnel_parm *p, __u32 fwmark);
void ip_tunnel_setup(struct net_device *dev, unsigned 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 */
f35581d6	61	#define IP_TUNNEL_INFO_BRIDGE 0x04 /* represents a bridged tunnel id */
ee122c79	62
fca5fdf6 DB	63	/* Maximum tunnel options length. */
	64	#define IP_TUNNEL_OPTS_MAX \
	65	GENMASK((FIELD_SIZEOF(struct ip_tunnel_info, \
	66	options_len) * BITS_PER_BYTE) - 1, 0)
	67
1d8fff90 TG	68	struct ip_tunnel_info {
1d8fff90 TG	69	struct ip_tunnel_key key;
d71785ff PA	70	#ifdef CONFIG_DST_CACHE
	71	struct dst_cache dst_cache;
	72	#endif
1d8fff90	73	u8 options_len;
ee122c79	74	u8 mode;
1d8fff90 TG	75	};
1d8fff90 TG	76
c5441932 PS	77	/* 6rd prefix/relay information */
	78	#ifdef CONFIG_IPV6_SIT_6RD
	79	struct ip_tunnel_6rd_parm {
	80	struct in6_addr prefix;
	81	__be32 relay_prefix;
	82	u16 prefixlen;
	83	u16 relay_prefixlen;
	84	};
	85	#endif
	86
56328486	87	struct ip_tunnel_encap {
6b8847c5 JB	88	u16 type;
6b8847c5 JB	89	u16 flags;
56328486 TH	90	__be16 sport;
	91	__be16 dport;
	92	};
	93
c5441932 PS	94	struct ip_tunnel_prl_entry {
	95	struct ip_tunnel_prl_entry __rcu *next;
	96	__be32 addr;
	97	u16 flags;
	98	struct rcu_head rcu_head;
	99	};
	100
2e15ea39 PS	101	struct metadata_dst;
2e15ea39 PS	102
c5441932 PS	103	struct ip_tunnel {
	104	struct ip_tunnel __rcu *next;
	105	struct hlist_node hash_node;
	106	struct net_device *dev;
5e6700b3	107	struct net net; / netns for packet i/o */
c5441932	108
c5441932 PS	109	unsigned long err_time; /* Time when the last ICMP error
c5441932 PS	110	* arrived */
f38ba953	111	int err_count; /* Number of arrived ICMP errors */
c5441932 PS	112
c5441932 PS	113	/* These four fields used only by GRE */
6b8847c5 JB	114	u32 i_seqno; /* The last seen seqno */
6b8847c5 JB	115	u32 o_seqno; /* The last output seqno */
56328486	116	int tun_hlen; /* Precalculated header length */
c5441932	117
84e54fe0 WT	118	/* This field used only by ERSPAN */
	119	u32 index; /* ERSPAN type II index */
	120
e09acddf	121	struct dst_cache dst_cache;
7d442fab	122
c5441932 PS	123	struct ip_tunnel_parm parms;
c5441932 PS	124
f38ba953	125	int mlink;
56328486	126	int encap_hlen; /* Encap header length (FOU,GUE) */
56328486	127	int hlen; /* tun_hlen + encap_hlen */
f38ba953	128	struct ip_tunnel_encap encap;
56328486	129
c5441932 PS	130	/* for SIT */
	131	#ifdef CONFIG_IPV6_SIT_6RD
	132	struct ip_tunnel_6rd_parm ip6rd;
	133	#endif
	134	struct ip_tunnel_prl_entry __rcu prl; / potential router list */
	135	unsigned int prl_count; /* # of entries in PRL */
c7d03a00	136	unsigned int ip_tnl_net_id;
c5441932	137	struct gro_cells gro_cells;
9830ad4c	138	__u32 fwmark;
2e15ea39	139	bool collect_md;
22a59be8	140	bool ignore_df;
c5441932 PS	141	};
c5441932 PS	142
f5796684 JG	143	#define TUNNEL_CSUM __cpu_to_be16(0x01)
	144	#define TUNNEL_ROUTING __cpu_to_be16(0x02)
	145	#define TUNNEL_KEY __cpu_to_be16(0x04)
	146	#define TUNNEL_SEQ __cpu_to_be16(0x08)
	147	#define TUNNEL_STRICT __cpu_to_be16(0x10)
	148	#define TUNNEL_REC __cpu_to_be16(0x20)
	149	#define TUNNEL_VERSION __cpu_to_be16(0x40)
	150	#define TUNNEL_NO_KEY __cpu_to_be16(0x80)
9a628224	151	#define TUNNEL_DONT_FRAGMENT __cpu_to_be16(0x0100)
f5796684 JG	152	#define TUNNEL_OAM __cpu_to_be16(0x0200)
f5796684 JG	153	#define TUNNEL_CRIT_OPT __cpu_to_be16(0x0400)
1dd144cf TG	154	#define TUNNEL_GENEVE_OPT __cpu_to_be16(0x0800)
1dd144cf TG	155	#define TUNNEL_VXLAN_OPT __cpu_to_be16(0x1000)
db3c6139	156	#define TUNNEL_NOCACHE __cpu_to_be16(0x2000)
1a66a836	157	#define TUNNEL_ERSPAN_OPT __cpu_to_be16(0x4000)
1dd144cf	158
1a66a836 WT	159	#define TUNNEL_OPTIONS_PRESENT \
1a66a836 WT	160	(TUNNEL_GENEVE_OPT \| TUNNEL_VXLAN_OPT \| TUNNEL_ERSPAN_OPT)
c5441932 PS	161
	162	struct tnl_ptk_info {
	163	__be16 flags;
	164	__be16 proto;
	165	__be32 key;
	166	__be32 seq;
9b8c6d7b	167	int hdr_len;
c5441932 PS	168	};
	169
	170	#define PACKET_RCVD 0
	171	#define PACKET_REJECT 1
125372fa	172	#define PACKET_NEXT 2
c5441932	173
6261d983	174	#define IP_TNL_HASH_BITS 7
c5441932 PS	175	#define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS)
	176
	177	struct ip_tunnel_net {
c5441932	178	struct net_device *fb_tunnel_dev;
6261d983	179	struct hlist_head tunnels[IP_TNL_HASH_SIZE];
2e15ea39	180	struct ip_tunnel __rcu *collect_md_tun;
c5441932 PS	181	};
c5441932 PS	182
4c222798 PS	183	static inline void ip_tunnel_key_init(struct ip_tunnel_key *key,
4c222798 PS	184	__be32 saddr, __be32 daddr,
13461144	185	u8 tos, u8 ttl, __be32 label,
4c222798 PS	186	__be16 tp_src, __be16 tp_dst,
4c222798 PS	187	__be64 tun_id, __be16 tun_flags)
1d8fff90	188	{
4c222798 PS	189	key->tun_id = tun_id;
	190	key->u.ipv4.src = saddr;
	191	key->u.ipv4.dst = daddr;
	192	memset((unsigned char *)key + IP_TUNNEL_KEY_IPV4_PAD,
c1ea5d67	193	0, IP_TUNNEL_KEY_IPV4_PAD_LEN);
4c222798 PS	194	key->tos = tos;
4c222798 PS	195	key->ttl = ttl;
13461144	196	key->label = label;
4c222798	197	key->tun_flags = tun_flags;
1d8fff90 TG	198
	199	/* For the tunnel types on the top of IPsec, the tp_src and tp_dst of
	200	* the upper tunnel are used.
	201	* E.g: GRE over IPSEC, the tp_src and tp_port are zero.
	202	*/
4c222798 PS	203	key->tp_src = tp_src;
4c222798 PS	204	key->tp_dst = tp_dst;
1d8fff90 TG	205
1d8fff90 TG	206	/* Clear struct padding. */
4c222798 PS	207	if (sizeof(*key) != IP_TUNNEL_KEY_SIZE)
	208	memset((unsigned char *)key + IP_TUNNEL_KEY_SIZE,
	209	0, sizeof(*key) - IP_TUNNEL_KEY_SIZE);
1d8fff90 TG	210	}
1d8fff90 TG	211
db3c6139 DB	212	static inline bool
	213	ip_tunnel_dst_cache_usable(const struct sk_buff *skb,
	214	const struct ip_tunnel_info *info)
	215	{
	216	if (skb->mark)
	217	return false;
	218	if (!info)
	219	return true;
	220	if (info->key.tun_flags & TUNNEL_NOCACHE)
	221	return false;
	222
	223	return true;
	224	}
	225
7f9562a1 JB	226	static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info
	227	*tun_info)
	228	{
	229	return tun_info->mode & IP_TUNNEL_INFO_IPV6 ? AF_INET6 : AF_INET;
	230	}
	231
d817f432 AV	232	static inline __be64 key32_to_tunnel_id(__be32 key)
	233	{
	234	#ifdef __BIG_ENDIAN
	235	return (__force __be64)key;
	236	#else
	237	return (__force __be64)((__force u64)key << 32);
	238	#endif
	239	}
	240
	241	/* Returns the least-significant 32 bits of a __be64. */
	242	static inline __be32 tunnel_id_to_key32(__be64 tun_id)
	243	{
	244	#ifdef __BIG_ENDIAN
	245	return (__force __be32)tun_id;
	246	#else
	247	return (__force __be32)((__force u64)tun_id >> 32);
	248	#endif
	249	}
	250
5243b6ac JG	251	#ifdef CONFIG_INET
5243b6ac JG	252
c5441932 PS	253	int ip_tunnel_init(struct net_device *dev);
	254	void ip_tunnel_uninit(struct net_device *dev);
	255	void ip_tunnel_dellink(struct net_device dev, struct list_head head);
1728d4fa	256	struct net ip_tunnel_get_link_net(const struct net_device dev);
1e99584b	257	int ip_tunnel_get_iflink(const struct net_device *dev);
c7d03a00	258	int ip_tunnel_init_net(struct net *net, unsigned int ip_tnl_net_id,
d3b6f614	259	struct rtnl_link_ops ops, char devname);
c5441932	260
6c742e71	261	void ip_tunnel_delete_net(struct ip_tunnel_net itn, struct rtnl_link_ops ops);
c5441932 PS	262
c5441932 PS	263	void ip_tunnel_xmit(struct sk_buff skb, struct net_device dev,
bf3d6a8f	264	const struct iphdr *tnl_params, const u8 protocol);
cfc7381b AS	265	void ip_md_tunnel_xmit(struct sk_buff skb, struct net_device dev,
cfc7381b AS	266	const u8 proto);
c5441932	267	int ip_tunnel_ioctl(struct net_device dev, struct ip_tunnel_parm p, int cmd);
7e059158	268	int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict);
c5441932 PS	269	int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu);
c5441932 PS	270
bc1f4470	271	void ip_tunnel_get_stats64(struct net_device *dev,
bc1f4470	272	struct rtnl_link_stats64 *tot);
c5441932 PS	273	struct ip_tunnel ip_tunnel_lookup(struct ip_tunnel_net itn,
	274	int link, __be16 flags,
	275	__be32 remote, __be32 local,
	276	__be32 key);
	277
	278	int ip_tunnel_rcv(struct ip_tunnel tunnel, struct sk_buff skb,
2e15ea39 PS	279	const struct tnl_ptk_info tpi, struct metadata_dst tun_dst,
2e15ea39 PS	280	bool log_ecn_error);
c5441932	281	int ip_tunnel_changelink(struct net_device dev, struct nlattr tb[],
9830ad4c	282	struct ip_tunnel_parm *p, __u32 fwmark);
c5441932	283	int ip_tunnel_newlink(struct net_device dev, struct nlattr tb[],
9830ad4c	284	struct ip_tunnel_parm *p, __u32 fwmark);
c7d03a00	285	void ip_tunnel_setup(struct net_device *dev, unsigned int net_id);
55c2bc14 TH	286
	287	struct ip_tunnel_encap_ops {
	288	size_t (encap_hlen)(struct ip_tunnel_encap e);
	289	int (build_header)(struct sk_buff skb, struct ip_tunnel_encap *e,
	290	u8 protocol, struct flowi4 fl4);
	291	};
	292
	293	#define MAX_IPTUN_ENCAP_OPS 8
	294
	295	extern const struct ip_tunnel_encap_ops __rcu *
	296	iptun_encaps[MAX_IPTUN_ENCAP_OPS];
	297
	298	int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op,
	299	unsigned int num);
	300	int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
	301	unsigned int num);
	302
56328486 TH	303	int ip_tunnel_encap_setup(struct ip_tunnel *t,
56328486 TH	304	struct ip_tunnel_encap *ipencap);
c5441932	305
55c2bc14 TH	306	static inline int ip_encap_hlen(struct ip_tunnel_encap *e)
	307	{
	308	const struct ip_tunnel_encap_ops *ops;
	309	int hlen = -EINVAL;
	310
	311	if (e->type == TUNNEL_ENCAP_NONE)
	312	return 0;
	313
	314	if (e->type >= MAX_IPTUN_ENCAP_OPS)
	315	return -EINVAL;
	316
	317	rcu_read_lock();
	318	ops = rcu_dereference(iptun_encaps[e->type]);
	319	if (likely(ops && ops->encap_hlen))
	320	hlen = ops->encap_hlen(e);
	321	rcu_read_unlock();
	322
	323	return hlen;
	324	}
	325
	326	static inline int ip_tunnel_encap(struct sk_buff skb, struct ip_tunnel t,
	327	u8 protocol, struct flowi4 fl4)
	328	{
	329	const struct ip_tunnel_encap_ops *ops;
	330	int ret = -EINVAL;
	331
	332	if (t->encap.type == TUNNEL_ENCAP_NONE)
	333	return 0;
	334
	335	if (t->encap.type >= MAX_IPTUN_ENCAP_OPS)
	336	return -EINVAL;
	337
	338	rcu_read_lock();
	339	ops = rcu_dereference(iptun_encaps[t->encap.type]);
	340	if (likely(ops && ops->build_header))
	341	ret = ops->build_header(skb, &t->encap, protocol, fl4);
	342	rcu_read_unlock();
	343
	344	return ret;
	345	}
	346
c5441932 PS	347	/* Extract dsfield from inner protocol */
	348	static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph,
	349	const struct sk_buff *skb)
	350	{
	351	if (skb->protocol == htons(ETH_P_IP))
	352	return iph->tos;
	353	else if (skb->protocol == htons(ETH_P_IPV6))
	354	return ipv6_get_dsfield((const struct ipv6hdr *)iph);
	355	else
	356	return 0;
	357	}
	358
	359	/* Propogate ECN bits out */
	360	static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
	361	const struct sk_buff *skb)
	362	{
	363	u8 inner = ip_tunnel_get_dsfield(iph, skb);
	364
	365	return INET_ECN_encapsulate(tos, inner);
	366	}
	367
a6d5bbf3 JB	368	int __iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
	369	__be16 inner_proto, bool raw_proto, bool xnet);
	370
	371	static inline int iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
	372	__be16 inner_proto, bool xnet)
	373	{
	374	return __iptunnel_pull_header(skb, hdr_len, inner_proto, false, xnet);
	375	}
	376
039f5062 PS	377	void iptunnel_xmit(struct sock sk, struct rtable rt, struct sk_buff *skb,
	378	__be32 src, __be32 dst, u8 proto,
	379	u8 tos, u8 ttl, __be16 df, bool xnet);
63d008a4 JB	380	struct metadata_dst iptunnel_metadata_reply(struct metadata_dst md,
63d008a4 JB	381	gfp_t flags);
0e6fbc5b	382
aed069df	383	int iptunnel_handle_offloads(struct sk_buff *skb, int gso_type_mask);
2d26f0a3	384
a09a4c8d JG	385	static inline int iptunnel_pull_offloads(struct sk_buff *skb)
	386	{
	387	if (skb_is_gso(skb)) {
	388	int err;
	389
	390	err = skb_unclone(skb, GFP_ATOMIC);
	391	if (unlikely(err))
	392	return err;
	393	skb_shinfo(skb)->gso_type &= ~(NETIF_F_GSO_ENCAP_ALL >>
	394	NETIF_F_GSO_SHIFT);
	395	}
	396
	397	skb->encapsulation = 0;
	398	return 0;
	399	}
	400
039f5062	401	static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len)
c5441932	402	{
039f5062 PS	403	if (pkt_len > 0) {
039f5062 PS	404	struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats);
c5441932	405
c5441932	406	u64_stats_update_begin(&tstats->syncp);
039f5062	407	tstats->tx_bytes += pkt_len;
c5441932 PS	408	tstats->tx_packets++;
c5441932 PS	409	u64_stats_update_end(&tstats->syncp);
b4fe85f9	410	put_cpu_ptr(tstats);
c5441932	411	} else {
039f5062 PS	412	struct net_device_stats *err_stats = &dev->stats;
	413
	414	if (pkt_len < 0) {
	415	err_stats->tx_errors++;
	416	err_stats->tx_aborted_errors++;
	417	} else {
	418	err_stats->tx_dropped++;
	419	}
c5441932 PS	420	}
c5441932 PS	421	}
5243b6ac	422
4c222798	423	static inline void ip_tunnel_info_opts(struct ip_tunnel_info info)
ee122c79 TG	424	{
	425	return info + 1;
	426	}
	427
4c222798 PS	428	static inline void ip_tunnel_info_opts_get(void *to,
	429	const struct ip_tunnel_info *info)
	430	{
	431	memcpy(to, info + 1, info->options_len);
	432	}
	433
	434	static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
	435	const void *from, int len)
	436	{
	437	memcpy(ip_tunnel_info_opts(info), from, len);
	438	info->options_len = len;
	439	}
	440
3093fbe7 TG	441	static inline struct ip_tunnel_info lwt_tun_info(struct lwtunnel_state lwtstate)
	442	{
	443	return (struct ip_tunnel_info *)lwtstate->data;
	444	}
	445
e7030878 TG	446	extern struct static_key ip_tunnel_metadata_cnt;
	447
	448	/* Returns > 0 if metadata should be collected */
	449	static inline int ip_tunnel_collect_metadata(void)
	450	{
	451	return static_key_false(&ip_tunnel_metadata_cnt);
	452	}
	453
045a0fa0 TG	454	void __init ip_tunnel_core_init(void);
045a0fa0 TG	455
e7030878 TG	456	void ip_tunnel_need_metadata(void);
	457	void ip_tunnel_unneed_metadata(void);
	458
05283187 TG	459	#else /* CONFIG_INET */
	460
	461	static inline struct ip_tunnel_info lwt_tun_info(struct lwtunnel_state lwtstate)
	462	{
	463	return NULL;
	464	}
	465
	466	static inline void ip_tunnel_need_metadata(void)
	467	{
	468	}
	469
	470	static inline void ip_tunnel_unneed_metadata(void)
	471	{
	472	}
	473
e28e87ed DB	474	static inline void ip_tunnel_info_opts_get(void *to,
	475	const struct ip_tunnel_info *info)
	476	{
	477	}
	478
	479	static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
	480	const void *from, int len)
	481	{
	482	info->options_len = 0;
	483	}
	484
5243b6ac JG	485	#endif /* CONFIG_INET */
5243b6ac JG	486
c5441932	487	#endif /* __NET_IP_TUNNELS_H */