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> 7 #include <linux/socket.h> 8 #include <linux/types.h> 9 #include <linux/u64_stats_sync.h> 10 #include <net/dsfield.h> 11 #include <net/gro_cells.h> 12 #include <net/inet_ecn.h> 13 #include <net/netns/generic.h> 14 #include <net/rtnetlink.h> 15 #include <net/lwtunnel.h> 16 17 #if IS_ENABLED(CONFIG_IPV6) 18 #include <net/ipv6.h> 19 #include <net/ip6_fib.h> 20 #include <net/ip6_route.h> 21 #endif 22 23 /* Keep error state on tunnel for 30 sec */ 24 #define IPTUNNEL_ERR_TIMEO (30*HZ) 25 26 /* Used to memset ip_tunnel padding. */ 27 #define IP_TUNNEL_KEY_SIZE offsetofend(struct ip_tunnel_key, tp_dst) 28 29 /* Used to memset ipv4 address padding. */ 30 #define IP_TUNNEL_KEY_IPV4_PAD offsetofend(struct ip_tunnel_key, u.ipv4.dst) 31 #define IP_TUNNEL_KEY_IPV4_PAD_LEN \ 32 (FIELD_SIZEOF(struct ip_tunnel_key, u) - \ 33 FIELD_SIZEOF(struct ip_tunnel_key, u.ipv4)) 34 35 struct ip_tunnel_key { 36 __be64 tun_id; 37 union { 38 struct { 39 __be32 src; 40 __be32 dst; 41 } ipv4; 42 struct { 43 struct in6_addr src; 44 struct in6_addr dst; 45 } ipv6; 46 } u; 47 __be16 tun_flags; 48 u8 tos; /* TOS for IPv4, TC for IPv6 */ 49 u8 ttl; /* TTL for IPv4, HL for IPv6 */ 50 __be16 tp_src; 51 __be16 tp_dst; 52 }; 53 54 /* Flags for ip_tunnel_info mode. */ 55 #define IP_TUNNEL_INFO_TX 0x01 /* represents tx tunnel parameters */ 56 #define IP_TUNNEL_INFO_IPV6 0x02 /* key contains IPv6 addresses */ 57 58 struct ip_tunnel_info { 59 struct ip_tunnel_key key; 60 const void *options; 61 u8 options_len; 62 u8 mode; 63 }; 64 65 /* 6rd prefix/relay information */ 66 #ifdef CONFIG_IPV6_SIT_6RD 67 struct ip_tunnel_6rd_parm { 68 struct in6_addr prefix; 69 __be32 relay_prefix; 70 u16 prefixlen; 71 u16 relay_prefixlen; 72 }; 73 #endif 74 75 struct ip_tunnel_encap { 76 u16 type; 77 u16 flags; 78 __be16 sport; 79 __be16 dport; 80 }; 81 82 struct ip_tunnel_prl_entry { 83 struct ip_tunnel_prl_entry __rcu *next; 84 __be32 addr; 85 u16 flags; 86 struct rcu_head rcu_head; 87 }; 88 89 struct ip_tunnel_dst { 90 struct dst_entry __rcu *dst; 91 __be32 saddr; 92 }; 93 94 struct metadata_dst; 95 96 struct ip_tunnel { 97 struct ip_tunnel __rcu *next; 98 struct hlist_node hash_node; 99 struct net_device *dev; 100 struct net *net; /* netns for packet i/o */ 101 102 int err_count; /* Number of arrived ICMP errors */ 103 unsigned long err_time; /* Time when the last ICMP error 104 * arrived */ 105 106 /* These four fields used only by GRE */ 107 u32 i_seqno; /* The last seen seqno */ 108 u32 o_seqno; /* The last output seqno */ 109 int tun_hlen; /* Precalculated header length */ 110 int mlink; 111 112 struct ip_tunnel_dst __percpu *dst_cache; 113 114 struct ip_tunnel_parm parms; 115 116 int encap_hlen; /* Encap header length (FOU,GUE) */ 117 struct ip_tunnel_encap encap; 118 119 int hlen; /* tun_hlen + encap_hlen */ 120 121 /* for SIT */ 122 #ifdef CONFIG_IPV6_SIT_6RD 123 struct ip_tunnel_6rd_parm ip6rd; 124 #endif 125 struct ip_tunnel_prl_entry __rcu *prl; /* potential router list */ 126 unsigned int prl_count; /* # of entries in PRL */ 127 int ip_tnl_net_id; 128 struct gro_cells gro_cells; 129 bool collect_md; 130 }; 131 132 #define TUNNEL_CSUM __cpu_to_be16(0x01) 133 #define TUNNEL_ROUTING __cpu_to_be16(0x02) 134 #define TUNNEL_KEY __cpu_to_be16(0x04) 135 #define TUNNEL_SEQ __cpu_to_be16(0x08) 136 #define TUNNEL_STRICT __cpu_to_be16(0x10) 137 #define TUNNEL_REC __cpu_to_be16(0x20) 138 #define TUNNEL_VERSION __cpu_to_be16(0x40) 139 #define TUNNEL_NO_KEY __cpu_to_be16(0x80) 140 #define TUNNEL_DONT_FRAGMENT __cpu_to_be16(0x0100) 141 #define TUNNEL_OAM __cpu_to_be16(0x0200) 142 #define TUNNEL_CRIT_OPT __cpu_to_be16(0x0400) 143 #define TUNNEL_GENEVE_OPT __cpu_to_be16(0x0800) 144 #define TUNNEL_VXLAN_OPT __cpu_to_be16(0x1000) 145 146 #define TUNNEL_OPTIONS_PRESENT (TUNNEL_GENEVE_OPT | TUNNEL_VXLAN_OPT) 147 148 struct tnl_ptk_info { 149 __be16 flags; 150 __be16 proto; 151 __be32 key; 152 __be32 seq; 153 }; 154 155 #define PACKET_RCVD 0 156 #define PACKET_REJECT 1 157 158 #define IP_TNL_HASH_BITS 7 159 #define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS) 160 161 struct ip_tunnel_net { 162 struct net_device *fb_tunnel_dev; 163 struct hlist_head tunnels[IP_TNL_HASH_SIZE]; 164 struct ip_tunnel __rcu *collect_md_tun; 165 }; 166 167 struct ip_tunnel_encap_ops { 168 size_t (*encap_hlen)(struct ip_tunnel_encap *e); 169 int (*build_header)(struct sk_buff *skb, struct ip_tunnel_encap *e, 170 u8 *protocol, struct flowi4 *fl4); 171 }; 172 173 #define MAX_IPTUN_ENCAP_OPS 8 174 175 extern const struct ip_tunnel_encap_ops __rcu * 176 iptun_encaps[MAX_IPTUN_ENCAP_OPS]; 177 178 int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op, 179 unsigned int num); 180 int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op, 181 unsigned int num); 182 183 static inline void __ip_tunnel_info_init(struct ip_tunnel_info *tun_info, 184 __be32 saddr, __be32 daddr, 185 u8 tos, u8 ttl, 186 __be16 tp_src, __be16 tp_dst, 187 __be64 tun_id, __be16 tun_flags, 188 const void *opts, u8 opts_len) 189 { 190 tun_info->key.tun_id = tun_id; 191 tun_info->key.u.ipv4.src = saddr; 192 tun_info->key.u.ipv4.dst = daddr; 193 memset((unsigned char *)&tun_info->key + IP_TUNNEL_KEY_IPV4_PAD, 194 0, IP_TUNNEL_KEY_IPV4_PAD_LEN); 195 tun_info->key.tos = tos; 196 tun_info->key.ttl = ttl; 197 tun_info->key.tun_flags = tun_flags; 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 */ 203 tun_info->key.tp_src = tp_src; 204 tun_info->key.tp_dst = tp_dst; 205 206 /* Clear struct padding. */ 207 if (sizeof(tun_info->key) != IP_TUNNEL_KEY_SIZE) 208 memset((unsigned char *)&tun_info->key + IP_TUNNEL_KEY_SIZE, 209 0, sizeof(tun_info->key) - IP_TUNNEL_KEY_SIZE); 210 211 tun_info->options = opts; 212 tun_info->options_len = opts_len; 213 214 tun_info->mode = 0; 215 } 216 217 static inline void ip_tunnel_info_init(struct ip_tunnel_info *tun_info, 218 const struct iphdr *iph, 219 __be16 tp_src, __be16 tp_dst, 220 __be64 tun_id, __be16 tun_flags, 221 const void *opts, u8 opts_len) 222 { 223 __ip_tunnel_info_init(tun_info, iph->saddr, iph->daddr, 224 iph->tos, iph->ttl, tp_src, tp_dst, 225 tun_id, tun_flags, opts, opts_len); 226 } 227 228 static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info 229 *tun_info) 230 { 231 return tun_info->mode & IP_TUNNEL_INFO_IPV6 ? AF_INET6 : AF_INET; 232 } 233 234 #ifdef CONFIG_INET 235 236 int ip_tunnel_init(struct net_device *dev); 237 void ip_tunnel_uninit(struct net_device *dev); 238 void ip_tunnel_dellink(struct net_device *dev, struct list_head *head); 239 struct net *ip_tunnel_get_link_net(const struct net_device *dev); 240 int ip_tunnel_get_iflink(const struct net_device *dev); 241 int ip_tunnel_init_net(struct net *net, int ip_tnl_net_id, 242 struct rtnl_link_ops *ops, char *devname); 243 244 void ip_tunnel_delete_net(struct ip_tunnel_net *itn, struct rtnl_link_ops *ops); 245 246 void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev, 247 const struct iphdr *tnl_params, const u8 protocol); 248 int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd); 249 int ip_tunnel_encap(struct sk_buff *skb, struct ip_tunnel *t, 250 u8 *protocol, struct flowi4 *fl4); 251 int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu); 252 253 struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev, 254 struct rtnl_link_stats64 *tot); 255 struct ip_tunnel *ip_tunnel_lookup(struct ip_tunnel_net *itn, 256 int link, __be16 flags, 257 __be32 remote, __be32 local, 258 __be32 key); 259 260 int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb, 261 const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst, 262 bool log_ecn_error); 263 int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[], 264 struct ip_tunnel_parm *p); 265 int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[], 266 struct ip_tunnel_parm *p); 267 void ip_tunnel_setup(struct net_device *dev, int net_id); 268 void ip_tunnel_dst_reset_all(struct ip_tunnel *t); 269 int ip_tunnel_encap_setup(struct ip_tunnel *t, 270 struct ip_tunnel_encap *ipencap); 271 272 /* Extract dsfield from inner protocol */ 273 static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph, 274 const struct sk_buff *skb) 275 { 276 if (skb->protocol == htons(ETH_P_IP)) 277 return iph->tos; 278 else if (skb->protocol == htons(ETH_P_IPV6)) 279 return ipv6_get_dsfield((const struct ipv6hdr *)iph); 280 else 281 return 0; 282 } 283 284 /* Propogate ECN bits out */ 285 static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph, 286 const struct sk_buff *skb) 287 { 288 u8 inner = ip_tunnel_get_dsfield(iph, skb); 289 290 return INET_ECN_encapsulate(tos, inner); 291 } 292 293 int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto); 294 int iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb, 295 __be32 src, __be32 dst, u8 proto, 296 u8 tos, u8 ttl, __be16 df, bool xnet); 297 298 struct sk_buff *iptunnel_handle_offloads(struct sk_buff *skb, bool gre_csum, 299 int gso_type_mask); 300 301 static inline void iptunnel_xmit_stats(int err, 302 struct net_device_stats *err_stats, 303 struct pcpu_sw_netstats __percpu *stats) 304 { 305 if (err > 0) { 306 struct pcpu_sw_netstats *tstats = this_cpu_ptr(stats); 307 308 u64_stats_update_begin(&tstats->syncp); 309 tstats->tx_bytes += err; 310 tstats->tx_packets++; 311 u64_stats_update_end(&tstats->syncp); 312 } else if (err < 0) { 313 err_stats->tx_errors++; 314 err_stats->tx_aborted_errors++; 315 } else { 316 err_stats->tx_dropped++; 317 } 318 } 319 320 static inline void *ip_tunnel_info_opts(struct ip_tunnel_info *info, size_t n) 321 { 322 return info + 1; 323 } 324 325 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate) 326 { 327 return (struct ip_tunnel_info *)lwtstate->data; 328 } 329 330 extern struct static_key ip_tunnel_metadata_cnt; 331 332 /* Returns > 0 if metadata should be collected */ 333 static inline int ip_tunnel_collect_metadata(void) 334 { 335 return static_key_false(&ip_tunnel_metadata_cnt); 336 } 337 338 void __init ip_tunnel_core_init(void); 339 340 void ip_tunnel_need_metadata(void); 341 void ip_tunnel_unneed_metadata(void); 342 343 #else /* CONFIG_INET */ 344 345 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate) 346 { 347 return NULL; 348 } 349 350 static inline void ip_tunnel_need_metadata(void) 351 { 352 } 353 354 static inline void ip_tunnel_unneed_metadata(void) 355 { 356 } 357 358 #endif /* CONFIG_INET */ 359 360 #endif /* __NET_IP_TUNNELS_H */ 361