1 #ifndef __NET_IP_TUNNELS_H
2 #define __NET_IP_TUNNELS_H 1
4 #include <linux/if_tunnel.h>
5 #include <linux/netdevice.h>
6 #include <linux/skbuff.h>
7 #include <linux/types.h>
8 #include <linux/u64_stats_sync.h>
9 #include <net/dsfield.h>
10 #include <net/gro_cells.h>
11 #include <net/inet_ecn.h>
12 #include <net/netns/generic.h>
13 #include <net/rtnetlink.h>
14 #include <net/lwtunnel.h>
16 #if IS_ENABLED(CONFIG_IPV6)
18 #include <net/ip6_fib.h>
19 #include <net/ip6_route.h>
22 /* Keep error state on tunnel for 30 sec */
23 #define IPTUNNEL_ERR_TIMEO (30*HZ)
25 /* Used to memset ip_tunnel padding. */
26 #define IP_TUNNEL_KEY_SIZE offsetofend(struct ip_tunnel_key, tp_dst)
28 /* Used to memset ipv4 address padding. */
29 #define IP_TUNNEL_KEY_IPV4_PAD offsetofend(struct ip_tunnel_key, u.ipv4.dst)
30 #define IP_TUNNEL_KEY_IPV4_PAD_LEN \
31 (FIELD_SIZEOF(struct ip_tunnel_key, u) - \
32 FIELD_SIZEOF(struct ip_tunnel_key, u.ipv4))
34 struct ip_tunnel_key {
47 u8 tos; /* TOS for IPv4, TC for IPv6 */
48 u8 ttl; /* TTL for IPv4, HL for IPv6 */
53 /* Flags for ip_tunnel_info mode. */
54 #define IP_TUNNEL_INFO_TX 0x01 /* represents tx tunnel parameters */
56 struct ip_tunnel_info {
57 struct ip_tunnel_key key;
63 /* 6rd prefix/relay information */
64 #ifdef CONFIG_IPV6_SIT_6RD
65 struct ip_tunnel_6rd_parm {
66 struct in6_addr prefix;
73 struct ip_tunnel_encap {
80 struct ip_tunnel_prl_entry {
81 struct ip_tunnel_prl_entry __rcu *next;
84 struct rcu_head rcu_head;
87 struct ip_tunnel_dst {
88 struct dst_entry __rcu *dst;
95 struct ip_tunnel __rcu *next;
96 struct hlist_node hash_node;
97 struct net_device *dev;
98 struct net *net; /* netns for packet i/o */
100 int err_count; /* Number of arrived ICMP errors */
101 unsigned long err_time; /* Time when the last ICMP error
104 /* These four fields used only by GRE */
105 u32 i_seqno; /* The last seen seqno */
106 u32 o_seqno; /* The last output seqno */
107 int tun_hlen; /* Precalculated header length */
110 struct ip_tunnel_dst __percpu *dst_cache;
112 struct ip_tunnel_parm parms;
114 int encap_hlen; /* Encap header length (FOU,GUE) */
115 struct ip_tunnel_encap encap;
117 int hlen; /* tun_hlen + encap_hlen */
120 #ifdef CONFIG_IPV6_SIT_6RD
121 struct ip_tunnel_6rd_parm ip6rd;
123 struct ip_tunnel_prl_entry __rcu *prl; /* potential router list */
124 unsigned int prl_count; /* # of entries in PRL */
126 struct gro_cells gro_cells;
130 #define TUNNEL_CSUM __cpu_to_be16(0x01)
131 #define TUNNEL_ROUTING __cpu_to_be16(0x02)
132 #define TUNNEL_KEY __cpu_to_be16(0x04)
133 #define TUNNEL_SEQ __cpu_to_be16(0x08)
134 #define TUNNEL_STRICT __cpu_to_be16(0x10)
135 #define TUNNEL_REC __cpu_to_be16(0x20)
136 #define TUNNEL_VERSION __cpu_to_be16(0x40)
137 #define TUNNEL_NO_KEY __cpu_to_be16(0x80)
138 #define TUNNEL_DONT_FRAGMENT __cpu_to_be16(0x0100)
139 #define TUNNEL_OAM __cpu_to_be16(0x0200)
140 #define TUNNEL_CRIT_OPT __cpu_to_be16(0x0400)
141 #define TUNNEL_GENEVE_OPT __cpu_to_be16(0x0800)
142 #define TUNNEL_VXLAN_OPT __cpu_to_be16(0x1000)
144 #define TUNNEL_OPTIONS_PRESENT (TUNNEL_GENEVE_OPT | TUNNEL_VXLAN_OPT)
146 struct tnl_ptk_info {
153 #define PACKET_RCVD 0
154 #define PACKET_REJECT 1
156 #define IP_TNL_HASH_BITS 7
157 #define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS)
159 struct ip_tunnel_net {
160 struct net_device *fb_tunnel_dev;
161 struct hlist_head tunnels[IP_TNL_HASH_SIZE];
162 struct ip_tunnel __rcu *collect_md_tun;
165 struct ip_tunnel_encap_ops {
166 size_t (*encap_hlen)(struct ip_tunnel_encap *e);
167 int (*build_header)(struct sk_buff *skb, struct ip_tunnel_encap *e,
168 u8 *protocol, struct flowi4 *fl4);
171 #define MAX_IPTUN_ENCAP_OPS 8
173 extern const struct ip_tunnel_encap_ops __rcu *
174 iptun_encaps[MAX_IPTUN_ENCAP_OPS];
176 int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op,
178 int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
181 static inline void __ip_tunnel_info_init(struct ip_tunnel_info *tun_info,
182 __be32 saddr, __be32 daddr,
184 __be16 tp_src, __be16 tp_dst,
185 __be64 tun_id, __be16 tun_flags,
186 const void *opts, u8 opts_len)
188 tun_info->key.tun_id = tun_id;
189 tun_info->key.u.ipv4.src = saddr;
190 tun_info->key.u.ipv4.dst = daddr;
191 memset((unsigned char *)&tun_info->key + IP_TUNNEL_KEY_IPV4_PAD,
192 0, IP_TUNNEL_KEY_IPV4_PAD_LEN);
193 tun_info->key.tos = tos;
194 tun_info->key.ttl = ttl;
195 tun_info->key.tun_flags = tun_flags;
197 /* For the tunnel types on the top of IPsec, the tp_src and tp_dst of
198 * the upper tunnel are used.
199 * E.g: GRE over IPSEC, the tp_src and tp_port are zero.
201 tun_info->key.tp_src = tp_src;
202 tun_info->key.tp_dst = tp_dst;
204 /* Clear struct padding. */
205 if (sizeof(tun_info->key) != IP_TUNNEL_KEY_SIZE)
206 memset((unsigned char *)&tun_info->key + IP_TUNNEL_KEY_SIZE,
207 0, sizeof(tun_info->key) - IP_TUNNEL_KEY_SIZE);
209 tun_info->options = opts;
210 tun_info->options_len = opts_len;
213 static inline void ip_tunnel_info_init(struct ip_tunnel_info *tun_info,
214 const struct iphdr *iph,
215 __be16 tp_src, __be16 tp_dst,
216 __be64 tun_id, __be16 tun_flags,
217 const void *opts, u8 opts_len)
219 __ip_tunnel_info_init(tun_info, iph->saddr, iph->daddr,
220 iph->tos, iph->ttl, tp_src, tp_dst,
221 tun_id, tun_flags, opts, opts_len);
226 int ip_tunnel_init(struct net_device *dev);
227 void ip_tunnel_uninit(struct net_device *dev);
228 void ip_tunnel_dellink(struct net_device *dev, struct list_head *head);
229 struct net *ip_tunnel_get_link_net(const struct net_device *dev);
230 int ip_tunnel_get_iflink(const struct net_device *dev);
231 int ip_tunnel_init_net(struct net *net, int ip_tnl_net_id,
232 struct rtnl_link_ops *ops, char *devname);
234 void ip_tunnel_delete_net(struct ip_tunnel_net *itn, struct rtnl_link_ops *ops);
236 void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
237 const struct iphdr *tnl_params, const u8 protocol);
238 int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd);
239 int ip_tunnel_encap(struct sk_buff *skb, struct ip_tunnel *t,
240 u8 *protocol, struct flowi4 *fl4);
241 int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu);
243 struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev,
244 struct rtnl_link_stats64 *tot);
245 struct ip_tunnel *ip_tunnel_lookup(struct ip_tunnel_net *itn,
246 int link, __be16 flags,
247 __be32 remote, __be32 local,
250 int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb,
251 const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst,
253 int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[],
254 struct ip_tunnel_parm *p);
255 int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[],
256 struct ip_tunnel_parm *p);
257 void ip_tunnel_setup(struct net_device *dev, int net_id);
258 void ip_tunnel_dst_reset_all(struct ip_tunnel *t);
259 int ip_tunnel_encap_setup(struct ip_tunnel *t,
260 struct ip_tunnel_encap *ipencap);
262 /* Extract dsfield from inner protocol */
263 static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph,
264 const struct sk_buff *skb)
266 if (skb->protocol == htons(ETH_P_IP))
268 else if (skb->protocol == htons(ETH_P_IPV6))
269 return ipv6_get_dsfield((const struct ipv6hdr *)iph);
274 /* Propogate ECN bits out */
275 static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
276 const struct sk_buff *skb)
278 u8 inner = ip_tunnel_get_dsfield(iph, skb);
280 return INET_ECN_encapsulate(tos, inner);
283 int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto);
284 int iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
285 __be32 src, __be32 dst, u8 proto,
286 u8 tos, u8 ttl, __be16 df, bool xnet);
288 struct sk_buff *iptunnel_handle_offloads(struct sk_buff *skb, bool gre_csum,
291 static inline void iptunnel_xmit_stats(int err,
292 struct net_device_stats *err_stats,
293 struct pcpu_sw_netstats __percpu *stats)
296 struct pcpu_sw_netstats *tstats = this_cpu_ptr(stats);
298 u64_stats_update_begin(&tstats->syncp);
299 tstats->tx_bytes += err;
300 tstats->tx_packets++;
301 u64_stats_update_end(&tstats->syncp);
302 } else if (err < 0) {
303 err_stats->tx_errors++;
304 err_stats->tx_aborted_errors++;
306 err_stats->tx_dropped++;
310 static inline void *ip_tunnel_info_opts(struct ip_tunnel_info *info, size_t n)
315 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
317 return (struct ip_tunnel_info *)lwtstate->data;
320 extern struct static_key ip_tunnel_metadata_cnt;
322 /* Returns > 0 if metadata should be collected */
323 static inline int ip_tunnel_collect_metadata(void)
325 return static_key_false(&ip_tunnel_metadata_cnt);
328 void __init ip_tunnel_core_init(void);
330 void ip_tunnel_need_metadata(void);
331 void ip_tunnel_unneed_metadata(void);
333 #else /* CONFIG_INET */
335 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
340 static inline void ip_tunnel_need_metadata(void)
344 static inline void ip_tunnel_unneed_metadata(void)
348 #endif /* CONFIG_INET */
350 #endif /* __NET_IP_TUNNELS_H */