1 /* 2 * INET An implementation of the TCP/IP protocol suite for the LINUX 3 * operating system. INET is implemented using the BSD Socket 4 * interface as the means of communication with the user level. 5 * 6 * Definitions for the IP module. 7 * 8 * Version: @(#)ip.h 1.0.2 05/07/93 9 * 10 * Authors: Ross Biro 11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> 12 * Alan Cox, <gw4pts@gw4pts.ampr.org> 13 * 14 * Changes: 15 * Mike McLagan : Routing by source 16 * 17 * This program is free software; you can redistribute it and/or 18 * modify it under the terms of the GNU General Public License 19 * as published by the Free Software Foundation; either version 20 * 2 of the License, or (at your option) any later version. 21 */ 22 #ifndef _IP_H 23 #define _IP_H 24 25 #include <linux/types.h> 26 #include <linux/ip.h> 27 #include <linux/in.h> 28 #include <linux/skbuff.h> 29 30 #include <net/inet_sock.h> 31 #include <net/route.h> 32 #include <net/snmp.h> 33 #include <net/flow.h> 34 #include <net/flow_dissector.h> 35 36 struct sock; 37 38 struct inet_skb_parm { 39 struct ip_options opt; /* Compiled IP options */ 40 unsigned char flags; 41 42 #define IPSKB_FORWARDED BIT(0) 43 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1) 44 #define IPSKB_XFRM_TRANSFORMED BIT(2) 45 #define IPSKB_FRAG_COMPLETE BIT(3) 46 #define IPSKB_REROUTED BIT(4) 47 #define IPSKB_DOREDIRECT BIT(5) 48 #define IPSKB_FRAG_PMTU BIT(6) 49 50 u16 frag_max_size; 51 }; 52 53 static inline unsigned int ip_hdrlen(const struct sk_buff *skb) 54 { 55 return ip_hdr(skb)->ihl * 4; 56 } 57 58 struct ipcm_cookie { 59 __be32 addr; 60 int oif; 61 struct ip_options_rcu *opt; 62 __u8 tx_flags; 63 __u8 ttl; 64 __s16 tos; 65 char priority; 66 }; 67 68 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) 69 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) 70 71 struct ip_ra_chain { 72 struct ip_ra_chain __rcu *next; 73 struct sock *sk; 74 union { 75 void (*destructor)(struct sock *); 76 struct sock *saved_sk; 77 }; 78 struct rcu_head rcu; 79 }; 80 81 extern struct ip_ra_chain __rcu *ip_ra_chain; 82 83 /* IP flags. */ 84 #define IP_CE 0x8000 /* Flag: "Congestion" */ 85 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */ 86 #define IP_MF 0x2000 /* Flag: "More Fragments" */ 87 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ 88 89 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ 90 91 struct msghdr; 92 struct net_device; 93 struct packet_type; 94 struct rtable; 95 struct sockaddr; 96 97 int igmp_mc_init(void); 98 99 /* 100 * Functions provided by ip.c 101 */ 102 103 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk, 104 __be32 saddr, __be32 daddr, 105 struct ip_options_rcu *opt); 106 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, 107 struct net_device *orig_dev); 108 int ip_local_deliver(struct sk_buff *skb); 109 int ip_mr_input(struct sk_buff *skb); 110 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb); 111 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb); 112 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, 113 int (*output)(struct net *, struct sock *, struct sk_buff *)); 114 void ip_send_check(struct iphdr *ip); 115 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 116 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 117 118 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); 119 void ip_init(void); 120 int ip_append_data(struct sock *sk, struct flowi4 *fl4, 121 int getfrag(void *from, char *to, int offset, int len, 122 int odd, struct sk_buff *skb), 123 void *from, int len, int protolen, 124 struct ipcm_cookie *ipc, 125 struct rtable **rt, 126 unsigned int flags); 127 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, 128 struct sk_buff *skb); 129 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page, 130 int offset, size_t size, int flags); 131 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, 132 struct sk_buff_head *queue, 133 struct inet_cork *cork); 134 int ip_send_skb(struct net *net, struct sk_buff *skb); 135 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); 136 void ip_flush_pending_frames(struct sock *sk); 137 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, 138 int getfrag(void *from, char *to, int offset, 139 int len, int odd, struct sk_buff *skb), 140 void *from, int length, int transhdrlen, 141 struct ipcm_cookie *ipc, struct rtable **rtp, 142 unsigned int flags); 143 144 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) 145 { 146 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base); 147 } 148 149 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet) 150 { 151 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos); 152 } 153 154 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk) 155 { 156 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk); 157 } 158 159 /* datagram.c */ 160 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 161 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 162 163 void ip4_datagram_release_cb(struct sock *sk); 164 165 struct ip_reply_arg { 166 struct kvec iov[1]; 167 int flags; 168 __wsum csum; 169 int csumoffset; /* u16 offset of csum in iov[0].iov_base */ 170 /* -1 if not needed */ 171 int bound_dev_if; 172 u8 tos; 173 }; 174 175 #define IP_REPLY_ARG_NOSRCCHECK 1 176 177 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg) 178 { 179 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0; 180 } 181 182 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, 183 const struct ip_options *sopt, 184 __be32 daddr, __be32 saddr, 185 const struct ip_reply_arg *arg, 186 unsigned int len); 187 188 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) 189 #define IP_INC_STATS_BH(net, field) SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field) 190 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 191 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val) 192 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 193 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val) 194 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) 195 #define NET_INC_STATS_BH(net, field) SNMP_INC_STATS_BH((net)->mib.net_statistics, field) 196 #define NET_INC_STATS_USER(net, field) SNMP_INC_STATS_USER((net)->mib.net_statistics, field) 197 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 198 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd) 199 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd) 200 201 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct); 202 unsigned long snmp_fold_field(void __percpu *mib, int offt); 203 #if BITS_PER_LONG==32 204 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 205 size_t syncp_offset); 206 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); 207 #else 208 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 209 size_t syncp_offset) 210 { 211 return snmp_get_cpu_field(mib, cpu, offct); 212 213 } 214 215 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) 216 { 217 return snmp_fold_field(mib, offt); 218 } 219 #endif 220 221 void inet_get_local_port_range(struct net *net, int *low, int *high); 222 223 #ifdef CONFIG_SYSCTL 224 static inline int inet_is_local_reserved_port(struct net *net, int port) 225 { 226 if (!net->ipv4.sysctl_local_reserved_ports) 227 return 0; 228 return test_bit(port, net->ipv4.sysctl_local_reserved_ports); 229 } 230 231 static inline bool sysctl_dev_name_is_allowed(const char *name) 232 { 233 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; 234 } 235 236 #else 237 static inline int inet_is_local_reserved_port(struct net *net, int port) 238 { 239 return 0; 240 } 241 #endif 242 243 __be32 inet_current_timestamp(void); 244 245 /* From inetpeer.c */ 246 extern int inet_peer_threshold; 247 extern int inet_peer_minttl; 248 extern int inet_peer_maxttl; 249 250 void ipfrag_init(void); 251 252 void ip_static_sysctl_init(void); 253 254 #define IP4_REPLY_MARK(net, mark) \ 255 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0) 256 257 static inline bool ip_is_fragment(const struct iphdr *iph) 258 { 259 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; 260 } 261 262 #ifdef CONFIG_INET 263 #include <net/dst.h> 264 265 /* The function in 2.2 was invalid, producing wrong result for 266 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ 267 static inline 268 int ip_decrease_ttl(struct iphdr *iph) 269 { 270 u32 check = (__force u32)iph->check; 271 check += (__force u32)htons(0x0100); 272 iph->check = (__force __sum16)(check + (check>=0xFFFF)); 273 return --iph->ttl; 274 } 275 276 static inline 277 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst) 278 { 279 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); 280 281 return pmtudisc == IP_PMTUDISC_DO || 282 (pmtudisc == IP_PMTUDISC_WANT && 283 !(dst_metric_locked(dst, RTAX_MTU))); 284 } 285 286 static inline bool ip_sk_accept_pmtu(const struct sock *sk) 287 { 288 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE && 289 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT; 290 } 291 292 static inline bool ip_sk_use_pmtu(const struct sock *sk) 293 { 294 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE; 295 } 296 297 static inline bool ip_sk_ignore_df(const struct sock *sk) 298 { 299 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO || 300 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT; 301 } 302 303 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, 304 bool forwarding) 305 { 306 struct net *net = dev_net(dst->dev); 307 308 if (net->ipv4.sysctl_ip_fwd_use_pmtu || 309 dst_metric_locked(dst, RTAX_MTU) || 310 !forwarding) 311 return dst_mtu(dst); 312 313 return min(dst->dev->mtu, IP_MAX_MTU); 314 } 315 316 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff *skb) 317 { 318 struct sock *sk = skb->sk; 319 320 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) { 321 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; 322 323 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding); 324 } 325 326 return min(skb_dst(skb)->dev->mtu, IP_MAX_MTU); 327 } 328 329 u32 ip_idents_reserve(u32 hash, int segs); 330 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); 331 332 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, 333 struct sock *sk, int segs) 334 { 335 struct iphdr *iph = ip_hdr(skb); 336 337 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { 338 /* This is only to work around buggy Windows95/2000 339 * VJ compression implementations. If the ID field 340 * does not change, they drop every other packet in 341 * a TCP stream using header compression. 342 */ 343 if (sk && inet_sk(sk)->inet_daddr) { 344 iph->id = htons(inet_sk(sk)->inet_id); 345 inet_sk(sk)->inet_id += segs; 346 } else { 347 iph->id = 0; 348 } 349 } else { 350 __ip_select_ident(net, iph, segs); 351 } 352 } 353 354 static inline void ip_select_ident(struct net *net, struct sk_buff *skb, 355 struct sock *sk) 356 { 357 ip_select_ident_segs(net, skb, sk, 1); 358 } 359 360 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) 361 { 362 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, 363 skb->len, proto, 0); 364 } 365 366 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store 367 * Equivalent to : flow->v4addrs.src = iph->saddr; 368 * flow->v4addrs.dst = iph->daddr; 369 */ 370 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, 371 const struct iphdr *iph) 372 { 373 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != 374 offsetof(typeof(flow->addrs), v4addrs.src) + 375 sizeof(flow->addrs.v4addrs.src)); 376 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs)); 377 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; 378 } 379 380 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto) 381 { 382 const struct iphdr *iph = skb_gro_network_header(skb); 383 384 return csum_tcpudp_nofold(iph->saddr, iph->daddr, 385 skb_gro_len(skb), proto, 0); 386 } 387 388 /* 389 * Map a multicast IP onto multicast MAC for type ethernet. 390 */ 391 392 static inline void ip_eth_mc_map(__be32 naddr, char *buf) 393 { 394 __u32 addr=ntohl(naddr); 395 buf[0]=0x01; 396 buf[1]=0x00; 397 buf[2]=0x5e; 398 buf[5]=addr&0xFF; 399 addr>>=8; 400 buf[4]=addr&0xFF; 401 addr>>=8; 402 buf[3]=addr&0x7F; 403 } 404 405 /* 406 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. 407 * Leave P_Key as 0 to be filled in by driver. 408 */ 409 410 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 411 { 412 __u32 addr; 413 unsigned char scope = broadcast[5] & 0xF; 414 415 buf[0] = 0; /* Reserved */ 416 buf[1] = 0xff; /* Multicast QPN */ 417 buf[2] = 0xff; 418 buf[3] = 0xff; 419 addr = ntohl(naddr); 420 buf[4] = 0xff; 421 buf[5] = 0x10 | scope; /* scope from broadcast address */ 422 buf[6] = 0x40; /* IPv4 signature */ 423 buf[7] = 0x1b; 424 buf[8] = broadcast[8]; /* P_Key */ 425 buf[9] = broadcast[9]; 426 buf[10] = 0; 427 buf[11] = 0; 428 buf[12] = 0; 429 buf[13] = 0; 430 buf[14] = 0; 431 buf[15] = 0; 432 buf[19] = addr & 0xff; 433 addr >>= 8; 434 buf[18] = addr & 0xff; 435 addr >>= 8; 436 buf[17] = addr & 0xff; 437 addr >>= 8; 438 buf[16] = addr & 0x0f; 439 } 440 441 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 442 { 443 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) 444 memcpy(buf, broadcast, 4); 445 else 446 memcpy(buf, &naddr, sizeof(naddr)); 447 } 448 449 #if IS_ENABLED(CONFIG_IPV6) 450 #include <linux/ipv6.h> 451 #endif 452 453 static __inline__ void inet_reset_saddr(struct sock *sk) 454 { 455 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; 456 #if IS_ENABLED(CONFIG_IPV6) 457 if (sk->sk_family == PF_INET6) { 458 struct ipv6_pinfo *np = inet6_sk(sk); 459 460 memset(&np->saddr, 0, sizeof(np->saddr)); 461 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); 462 } 463 #endif 464 } 465 466 #endif 467 468 static inline unsigned int ipv4_addr_hash(__be32 ip) 469 { 470 return (__force unsigned int) ip; 471 } 472 473 bool ip_call_ra_chain(struct sk_buff *skb); 474 475 /* 476 * Functions provided by ip_fragment.c 477 */ 478 479 enum ip_defrag_users { 480 IP_DEFRAG_LOCAL_DELIVER, 481 IP_DEFRAG_CALL_RA_CHAIN, 482 IP_DEFRAG_CONNTRACK_IN, 483 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, 484 IP_DEFRAG_CONNTRACK_OUT, 485 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 486 IP_DEFRAG_CONNTRACK_BRIDGE_IN, 487 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 488 IP_DEFRAG_VS_IN, 489 IP_DEFRAG_VS_OUT, 490 IP_DEFRAG_VS_FWD, 491 IP_DEFRAG_AF_PACKET, 492 IP_DEFRAG_MACVLAN, 493 }; 494 495 /* Return true if the value of 'user' is between 'lower_bond' 496 * and 'upper_bond' inclusively. 497 */ 498 static inline bool ip_defrag_user_in_between(u32 user, 499 enum ip_defrag_users lower_bond, 500 enum ip_defrag_users upper_bond) 501 { 502 return user >= lower_bond && user <= upper_bond; 503 } 504 505 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user); 506 #ifdef CONFIG_INET 507 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user); 508 #else 509 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user) 510 { 511 return skb; 512 } 513 #endif 514 int ip_frag_mem(struct net *net); 515 516 /* 517 * Functions provided by ip_forward.c 518 */ 519 520 int ip_forward(struct sk_buff *skb); 521 522 /* 523 * Functions provided by ip_options.c 524 */ 525 526 void ip_options_build(struct sk_buff *skb, struct ip_options *opt, 527 __be32 daddr, struct rtable *rt, int is_frag); 528 529 int __ip_options_echo(struct ip_options *dopt, struct sk_buff *skb, 530 const struct ip_options *sopt); 531 static inline int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb) 532 { 533 return __ip_options_echo(dopt, skb, &IPCB(skb)->opt); 534 } 535 536 void ip_options_fragment(struct sk_buff *skb); 537 int ip_options_compile(struct net *net, struct ip_options *opt, 538 struct sk_buff *skb); 539 int ip_options_get(struct net *net, struct ip_options_rcu **optp, 540 unsigned char *data, int optlen); 541 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp, 542 unsigned char __user *data, int optlen); 543 void ip_options_undo(struct ip_options *opt); 544 void ip_forward_options(struct sk_buff *skb); 545 int ip_options_rcv_srr(struct sk_buff *skb); 546 547 /* 548 * Functions provided by ip_sockglue.c 549 */ 550 551 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb); 552 void ip_cmsg_recv_offset(struct msghdr *msg, struct sk_buff *skb, int offset); 553 int ip_cmsg_send(struct net *net, struct msghdr *msg, 554 struct ipcm_cookie *ipc, bool allow_ipv6); 555 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, 556 unsigned int optlen); 557 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, 558 int __user *optlen); 559 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 560 char __user *optval, unsigned int optlen); 561 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 562 char __user *optval, int __user *optlen); 563 int ip_ra_control(struct sock *sk, unsigned char on, 564 void (*destructor)(struct sock *)); 565 566 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); 567 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 568 u32 info, u8 *payload); 569 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, 570 u32 info); 571 572 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 573 { 574 ip_cmsg_recv_offset(msg, skb, 0); 575 } 576 577 bool icmp_global_allow(void); 578 extern int sysctl_icmp_msgs_per_sec; 579 extern int sysctl_icmp_msgs_burst; 580 581 #ifdef CONFIG_PROC_FS 582 int ip_misc_proc_init(void); 583 #endif 584 585 #endif /* _IP_H */ 586