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/socket.h>
8 #include <linux/types.h>
9 #include <linux/u64_stats_sync.h>
10 #include <linux/bitops.h>
12 #include <net/dsfield.h>
13 #include <net/gro_cells.h>
14 #include <net/inet_ecn.h>
15 #include <net/netns/generic.h>
16 #include <net/rtnetlink.h>
17 #include <net/lwtunnel.h>
18 #include <net/dst_cache.h>
20 #if IS_ENABLED(CONFIG_IPV6)
22 #include <net/ip6_fib.h>
23 #include <net/ip6_route.h>
26 /* Keep error state on tunnel for 30 sec */
27 #define IPTUNNEL_ERR_TIMEO (30*HZ)
29 /* Used to memset ip_tunnel padding. */
30 #define IP_TUNNEL_KEY_SIZE offsetofend(struct ip_tunnel_key, tp_dst)
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))
38 struct ip_tunnel_key {
51 u8 tos; /* TOS for IPv4, TC for IPv6 */
52 u8 ttl; /* TTL for IPv4, HL for IPv6 */
53 __be32 label; /* Flow Label for IPv6 */
58 /* Flags for ip_tunnel_info mode. */
59 #define IP_TUNNEL_INFO_TX 0x01 /* represents tx tunnel parameters */
60 #define IP_TUNNEL_INFO_IPV6 0x02 /* key contains IPv6 addresses */
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)
67 struct ip_tunnel_info {
68 struct ip_tunnel_key key;
69 #ifdef CONFIG_DST_CACHE
70 struct dst_cache dst_cache;
76 /* 6rd prefix/relay information */
77 #ifdef CONFIG_IPV6_SIT_6RD
78 struct ip_tunnel_6rd_parm {
79 struct in6_addr prefix;
86 struct ip_tunnel_encap {
93 struct ip_tunnel_prl_entry {
94 struct ip_tunnel_prl_entry __rcu *next;
97 struct rcu_head rcu_head;
103 struct ip_tunnel __rcu *next;
104 struct hlist_node hash_node;
105 struct net_device *dev;
106 struct net *net; /* netns for packet i/o */
108 unsigned long err_time; /* Time when the last ICMP error
110 int err_count; /* Number of arrived ICMP errors */
112 /* These four fields used only by GRE */
113 u32 i_seqno; /* The last seen seqno */
114 u32 o_seqno; /* The last output seqno */
115 int tun_hlen; /* Precalculated header length */
117 struct dst_cache dst_cache;
119 struct ip_tunnel_parm parms;
122 int encap_hlen; /* Encap header length (FOU,GUE) */
123 int hlen; /* tun_hlen + encap_hlen */
124 struct ip_tunnel_encap encap;
127 #ifdef CONFIG_IPV6_SIT_6RD
128 struct ip_tunnel_6rd_parm ip6rd;
130 struct ip_tunnel_prl_entry __rcu *prl; /* potential router list */
131 unsigned int prl_count; /* # of entries in PRL */
133 struct gro_cells gro_cells;
137 #define TUNNEL_CSUM __cpu_to_be16(0x01)
138 #define TUNNEL_ROUTING __cpu_to_be16(0x02)
139 #define TUNNEL_KEY __cpu_to_be16(0x04)
140 #define TUNNEL_SEQ __cpu_to_be16(0x08)
141 #define TUNNEL_STRICT __cpu_to_be16(0x10)
142 #define TUNNEL_REC __cpu_to_be16(0x20)
143 #define TUNNEL_VERSION __cpu_to_be16(0x40)
144 #define TUNNEL_NO_KEY __cpu_to_be16(0x80)
145 #define TUNNEL_DONT_FRAGMENT __cpu_to_be16(0x0100)
146 #define TUNNEL_OAM __cpu_to_be16(0x0200)
147 #define TUNNEL_CRIT_OPT __cpu_to_be16(0x0400)
148 #define TUNNEL_GENEVE_OPT __cpu_to_be16(0x0800)
149 #define TUNNEL_VXLAN_OPT __cpu_to_be16(0x1000)
150 #define TUNNEL_NOCACHE __cpu_to_be16(0x2000)
152 #define TUNNEL_OPTIONS_PRESENT (TUNNEL_GENEVE_OPT | TUNNEL_VXLAN_OPT)
154 struct tnl_ptk_info {
161 #define PACKET_RCVD 0
162 #define PACKET_REJECT 1
164 #define IP_TNL_HASH_BITS 7
165 #define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS)
167 struct ip_tunnel_net {
168 struct net_device *fb_tunnel_dev;
169 struct hlist_head tunnels[IP_TNL_HASH_SIZE];
170 struct ip_tunnel __rcu *collect_md_tun;
173 struct ip_tunnel_encap_ops {
174 size_t (*encap_hlen)(struct ip_tunnel_encap *e);
175 int (*build_header)(struct sk_buff *skb, struct ip_tunnel_encap *e,
176 u8 *protocol, struct flowi4 *fl4);
179 #define MAX_IPTUN_ENCAP_OPS 8
181 extern const struct ip_tunnel_encap_ops __rcu *
182 iptun_encaps[MAX_IPTUN_ENCAP_OPS];
184 int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op,
186 int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
189 static inline void ip_tunnel_key_init(struct ip_tunnel_key *key,
190 __be32 saddr, __be32 daddr,
191 u8 tos, u8 ttl, __be32 label,
192 __be16 tp_src, __be16 tp_dst,
193 __be64 tun_id, __be16 tun_flags)
195 key->tun_id = tun_id;
196 key->u.ipv4.src = saddr;
197 key->u.ipv4.dst = daddr;
198 memset((unsigned char *)key + IP_TUNNEL_KEY_IPV4_PAD,
199 0, IP_TUNNEL_KEY_IPV4_PAD_LEN);
203 key->tun_flags = tun_flags;
205 /* For the tunnel types on the top of IPsec, the tp_src and tp_dst of
206 * the upper tunnel are used.
207 * E.g: GRE over IPSEC, the tp_src and tp_port are zero.
209 key->tp_src = tp_src;
210 key->tp_dst = tp_dst;
212 /* Clear struct padding. */
213 if (sizeof(*key) != IP_TUNNEL_KEY_SIZE)
214 memset((unsigned char *)key + IP_TUNNEL_KEY_SIZE,
215 0, sizeof(*key) - IP_TUNNEL_KEY_SIZE);
219 ip_tunnel_dst_cache_usable(const struct sk_buff *skb,
220 const struct ip_tunnel_info *info)
226 if (info->key.tun_flags & TUNNEL_NOCACHE)
232 static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info
235 return tun_info->mode & IP_TUNNEL_INFO_IPV6 ? AF_INET6 : AF_INET;
240 int ip_tunnel_init(struct net_device *dev);
241 void ip_tunnel_uninit(struct net_device *dev);
242 void ip_tunnel_dellink(struct net_device *dev, struct list_head *head);
243 struct net *ip_tunnel_get_link_net(const struct net_device *dev);
244 int ip_tunnel_get_iflink(const struct net_device *dev);
245 int ip_tunnel_init_net(struct net *net, int ip_tnl_net_id,
246 struct rtnl_link_ops *ops, char *devname);
248 void ip_tunnel_delete_net(struct ip_tunnel_net *itn, struct rtnl_link_ops *ops);
250 void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
251 const struct iphdr *tnl_params, const u8 protocol);
252 int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd);
253 int ip_tunnel_encap(struct sk_buff *skb, struct ip_tunnel *t,
254 u8 *protocol, struct flowi4 *fl4);
255 int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict);
256 int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu);
258 struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev,
259 struct rtnl_link_stats64 *tot);
260 struct ip_tunnel *ip_tunnel_lookup(struct ip_tunnel_net *itn,
261 int link, __be16 flags,
262 __be32 remote, __be32 local,
265 int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb,
266 const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst,
268 int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[],
269 struct ip_tunnel_parm *p);
270 int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[],
271 struct ip_tunnel_parm *p);
272 void ip_tunnel_setup(struct net_device *dev, int net_id);
273 int ip_tunnel_encap_setup(struct ip_tunnel *t,
274 struct ip_tunnel_encap *ipencap);
276 /* Extract dsfield from inner protocol */
277 static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph,
278 const struct sk_buff *skb)
280 if (skb->protocol == htons(ETH_P_IP))
282 else if (skb->protocol == htons(ETH_P_IPV6))
283 return ipv6_get_dsfield((const struct ipv6hdr *)iph);
288 /* Propogate ECN bits out */
289 static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
290 const struct sk_buff *skb)
292 u8 inner = ip_tunnel_get_dsfield(iph, skb);
294 return INET_ECN_encapsulate(tos, inner);
297 int __iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
298 __be16 inner_proto, bool raw_proto, bool xnet);
300 static inline int iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
301 __be16 inner_proto, bool xnet)
303 return __iptunnel_pull_header(skb, hdr_len, inner_proto, false, xnet);
306 void iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
307 __be32 src, __be32 dst, u8 proto,
308 u8 tos, u8 ttl, __be16 df, bool xnet);
309 struct metadata_dst *iptunnel_metadata_reply(struct metadata_dst *md,
312 int iptunnel_handle_offloads(struct sk_buff *skb, int gso_type_mask);
314 static inline int iptunnel_pull_offloads(struct sk_buff *skb)
316 if (skb_is_gso(skb)) {
319 err = skb_unclone(skb, GFP_ATOMIC);
322 skb_shinfo(skb)->gso_type &= ~(NETIF_F_GSO_ENCAP_ALL >>
326 skb->encapsulation = 0;
330 static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len)
333 struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats);
335 u64_stats_update_begin(&tstats->syncp);
336 tstats->tx_bytes += pkt_len;
337 tstats->tx_packets++;
338 u64_stats_update_end(&tstats->syncp);
341 struct net_device_stats *err_stats = &dev->stats;
344 err_stats->tx_errors++;
345 err_stats->tx_aborted_errors++;
347 err_stats->tx_dropped++;
352 static inline void *ip_tunnel_info_opts(struct ip_tunnel_info *info)
357 static inline void ip_tunnel_info_opts_get(void *to,
358 const struct ip_tunnel_info *info)
360 memcpy(to, info + 1, info->options_len);
363 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
364 const void *from, int len)
366 memcpy(ip_tunnel_info_opts(info), from, len);
367 info->options_len = len;
370 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
372 return (struct ip_tunnel_info *)lwtstate->data;
375 extern struct static_key ip_tunnel_metadata_cnt;
377 /* Returns > 0 if metadata should be collected */
378 static inline int ip_tunnel_collect_metadata(void)
380 return static_key_false(&ip_tunnel_metadata_cnt);
383 void __init ip_tunnel_core_init(void);
385 void ip_tunnel_need_metadata(void);
386 void ip_tunnel_unneed_metadata(void);
388 #else /* CONFIG_INET */
390 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
395 static inline void ip_tunnel_need_metadata(void)
399 static inline void ip_tunnel_unneed_metadata(void)
403 static inline void ip_tunnel_info_opts_get(void *to,
404 const struct ip_tunnel_info *info)
408 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
409 const void *from, int len)
411 info->options_len = 0;
414 #endif /* CONFIG_INET */
416 #endif /* __NET_IP_TUNNELS_H */