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 <linux/bitops.h> 11 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> 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 29 /* Used to memset ip_tunnel padding. */ 30 #define IP_TUNNEL_KEY_SIZE offsetofend(struct ip_tunnel_key, tp_dst) 31 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 38 struct ip_tunnel_key { 39 __be64 tun_id; 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; 50 __be16 tun_flags; 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 */ 54 __be16 tp_src; 55 __be16 tp_dst; 56 }; 57 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 */ 61 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 67 struct ip_tunnel_info { 68 struct ip_tunnel_key key; 69 #ifdef CONFIG_DST_CACHE 70 struct dst_cache dst_cache; 71 #endif 72 u8 options_len; 73 u8 mode; 74 }; 75 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 86 struct ip_tunnel_encap { 87 u16 type; 88 u16 flags; 89 __be16 sport; 90 __be16 dport; 91 }; 92 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 }; 99 100 struct metadata_dst; 101 102 struct ip_tunnel { 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 */ 107 108 int err_count; /* Number of arrived ICMP errors */ 109 unsigned long err_time; /* Time when the last ICMP error 110 * arrived */ 111 112 /* 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 */ 116 int mlink; 117 118 struct dst_cache dst_cache; 119 120 struct ip_tunnel_parm parms; 121 122 int encap_hlen; /* Encap header length (FOU,GUE) */ 123 struct ip_tunnel_encap encap; 124 125 int hlen; /* tun_hlen + encap_hlen */ 126 127 /* for SIT */ 128 #ifdef CONFIG_IPV6_SIT_6RD 129 struct ip_tunnel_6rd_parm ip6rd; 130 #endif 131 struct ip_tunnel_prl_entry __rcu *prl; /* potential router list */ 132 unsigned int prl_count; /* # of entries in PRL */ 133 int ip_tnl_net_id; 134 struct gro_cells gro_cells; 135 bool collect_md; 136 }; 137 138 #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) 146 #define TUNNEL_DONT_FRAGMENT __cpu_to_be16(0x0100) 147 #define TUNNEL_OAM __cpu_to_be16(0x0200) 148 #define TUNNEL_CRIT_OPT __cpu_to_be16(0x0400) 149 #define TUNNEL_GENEVE_OPT __cpu_to_be16(0x0800) 150 #define TUNNEL_VXLAN_OPT __cpu_to_be16(0x1000) 151 #define TUNNEL_NOCACHE __cpu_to_be16(0x2000) 152 153 #define TUNNEL_OPTIONS_PRESENT (TUNNEL_GENEVE_OPT | TUNNEL_VXLAN_OPT) 154 155 struct tnl_ptk_info { 156 __be16 flags; 157 __be16 proto; 158 __be32 key; 159 __be32 seq; 160 }; 161 162 #define PACKET_RCVD 0 163 #define PACKET_REJECT 1 164 165 #define IP_TNL_HASH_BITS 7 166 #define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS) 167 168 struct ip_tunnel_net { 169 struct net_device *fb_tunnel_dev; 170 struct hlist_head tunnels[IP_TNL_HASH_SIZE]; 171 struct ip_tunnel __rcu *collect_md_tun; 172 }; 173 174 struct ip_tunnel_encap_ops { 175 size_t (*encap_hlen)(struct ip_tunnel_encap *e); 176 int (*build_header)(struct sk_buff *skb, struct ip_tunnel_encap *e, 177 u8 *protocol, struct flowi4 *fl4); 178 }; 179 180 #define MAX_IPTUN_ENCAP_OPS 8 181 182 extern const struct ip_tunnel_encap_ops __rcu * 183 iptun_encaps[MAX_IPTUN_ENCAP_OPS]; 184 185 int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op, 186 unsigned int num); 187 int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op, 188 unsigned int num); 189 190 static inline void ip_tunnel_key_init(struct ip_tunnel_key *key, 191 __be32 saddr, __be32 daddr, 192 u8 tos, u8 ttl, __be32 label, 193 __be16 tp_src, __be16 tp_dst, 194 __be64 tun_id, __be16 tun_flags) 195 { 196 key->tun_id = tun_id; 197 key->u.ipv4.src = saddr; 198 key->u.ipv4.dst = daddr; 199 memset((unsigned char *)key + IP_TUNNEL_KEY_IPV4_PAD, 200 0, IP_TUNNEL_KEY_IPV4_PAD_LEN); 201 key->tos = tos; 202 key->ttl = ttl; 203 key->label = label; 204 key->tun_flags = tun_flags; 205 206 /* For the tunnel types on the top of IPsec, the tp_src and tp_dst of 207 * the upper tunnel are used. 208 * E.g: GRE over IPSEC, the tp_src and tp_port are zero. 209 */ 210 key->tp_src = tp_src; 211 key->tp_dst = tp_dst; 212 213 /* Clear struct padding. */ 214 if (sizeof(*key) != IP_TUNNEL_KEY_SIZE) 215 memset((unsigned char *)key + IP_TUNNEL_KEY_SIZE, 216 0, sizeof(*key) - IP_TUNNEL_KEY_SIZE); 217 } 218 219 static inline bool 220 ip_tunnel_dst_cache_usable(const struct sk_buff *skb, 221 const struct ip_tunnel_info *info) 222 { 223 if (skb->mark) 224 return false; 225 if (!info) 226 return true; 227 if (info->key.tun_flags & TUNNEL_NOCACHE) 228 return false; 229 230 return true; 231 } 232 233 static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info 234 *tun_info) 235 { 236 return tun_info->mode & IP_TUNNEL_INFO_IPV6 ? AF_INET6 : AF_INET; 237 } 238 239 #ifdef CONFIG_INET 240 241 int ip_tunnel_init(struct net_device *dev); 242 void ip_tunnel_uninit(struct net_device *dev); 243 void ip_tunnel_dellink(struct net_device *dev, struct list_head *head); 244 struct net *ip_tunnel_get_link_net(const struct net_device *dev); 245 int ip_tunnel_get_iflink(const struct net_device *dev); 246 int ip_tunnel_init_net(struct net *net, int ip_tnl_net_id, 247 struct rtnl_link_ops *ops, char *devname); 248 249 void ip_tunnel_delete_net(struct ip_tunnel_net *itn, struct rtnl_link_ops *ops); 250 251 void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev, 252 const struct iphdr *tnl_params, const u8 protocol); 253 int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd); 254 int ip_tunnel_encap(struct sk_buff *skb, struct ip_tunnel *t, 255 u8 *protocol, struct flowi4 *fl4); 256 int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict); 257 int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu); 258 259 struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev, 260 struct rtnl_link_stats64 *tot); 261 struct ip_tunnel *ip_tunnel_lookup(struct ip_tunnel_net *itn, 262 int link, __be16 flags, 263 __be32 remote, __be32 local, 264 __be32 key); 265 266 int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb, 267 const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst, 268 bool log_ecn_error); 269 int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[], 270 struct ip_tunnel_parm *p); 271 int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[], 272 struct ip_tunnel_parm *p); 273 void ip_tunnel_setup(struct net_device *dev, int net_id); 274 int ip_tunnel_encap_setup(struct ip_tunnel *t, 275 struct ip_tunnel_encap *ipencap); 276 277 /* Extract dsfield from inner protocol */ 278 static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph, 279 const struct sk_buff *skb) 280 { 281 if (skb->protocol == htons(ETH_P_IP)) 282 return iph->tos; 283 else if (skb->protocol == htons(ETH_P_IPV6)) 284 return ipv6_get_dsfield((const struct ipv6hdr *)iph); 285 else 286 return 0; 287 } 288 289 /* Propogate ECN bits out */ 290 static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph, 291 const struct sk_buff *skb) 292 { 293 u8 inner = ip_tunnel_get_dsfield(iph, skb); 294 295 return INET_ECN_encapsulate(tos, inner); 296 } 297 298 int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto, 299 bool xnet); 300 void iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb, 301 __be32 src, __be32 dst, u8 proto, 302 u8 tos, u8 ttl, __be16 df, bool xnet); 303 struct metadata_dst *iptunnel_metadata_reply(struct metadata_dst *md, 304 gfp_t flags); 305 306 struct sk_buff *iptunnel_handle_offloads(struct sk_buff *skb, int gso_type_mask); 307 308 static inline int iptunnel_pull_offloads(struct sk_buff *skb) 309 { 310 if (skb_is_gso(skb)) { 311 int err; 312 313 err = skb_unclone(skb, GFP_ATOMIC); 314 if (unlikely(err)) 315 return err; 316 skb_shinfo(skb)->gso_type &= ~(NETIF_F_GSO_ENCAP_ALL >> 317 NETIF_F_GSO_SHIFT); 318 } 319 320 skb->encapsulation = 0; 321 return 0; 322 } 323 324 static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len) 325 { 326 if (pkt_len > 0) { 327 struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats); 328 329 u64_stats_update_begin(&tstats->syncp); 330 tstats->tx_bytes += pkt_len; 331 tstats->tx_packets++; 332 u64_stats_update_end(&tstats->syncp); 333 put_cpu_ptr(tstats); 334 } else { 335 struct net_device_stats *err_stats = &dev->stats; 336 337 if (pkt_len < 0) { 338 err_stats->tx_errors++; 339 err_stats->tx_aborted_errors++; 340 } else { 341 err_stats->tx_dropped++; 342 } 343 } 344 } 345 346 static inline void *ip_tunnel_info_opts(struct ip_tunnel_info *info) 347 { 348 return info + 1; 349 } 350 351 static inline void ip_tunnel_info_opts_get(void *to, 352 const struct ip_tunnel_info *info) 353 { 354 memcpy(to, info + 1, info->options_len); 355 } 356 357 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info, 358 const void *from, int len) 359 { 360 memcpy(ip_tunnel_info_opts(info), from, len); 361 info->options_len = len; 362 } 363 364 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate) 365 { 366 return (struct ip_tunnel_info *)lwtstate->data; 367 } 368 369 extern struct static_key ip_tunnel_metadata_cnt; 370 371 /* Returns > 0 if metadata should be collected */ 372 static inline int ip_tunnel_collect_metadata(void) 373 { 374 return static_key_false(&ip_tunnel_metadata_cnt); 375 } 376 377 void __init ip_tunnel_core_init(void); 378 379 void ip_tunnel_need_metadata(void); 380 void ip_tunnel_unneed_metadata(void); 381 382 #else /* CONFIG_INET */ 383 384 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate) 385 { 386 return NULL; 387 } 388 389 static inline void ip_tunnel_need_metadata(void) 390 { 391 } 392 393 static inline void ip_tunnel_unneed_metadata(void) 394 { 395 } 396 397 static inline void ip_tunnel_info_opts_get(void *to, 398 const struct ip_tunnel_info *info) 399 { 400 } 401 402 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info, 403 const void *from, int len) 404 { 405 info->options_len = 0; 406 } 407 408 #endif /* CONFIG_INET */ 409 410 #endif /* __NET_IP_TUNNELS_H */ 411