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_keys.h> 35 36struct sock; 37 38struct 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 49 u16 frag_max_size; 50}; 51 52static inline unsigned int ip_hdrlen(const struct sk_buff *skb) 53{ 54 return ip_hdr(skb)->ihl * 4; 55} 56 57struct ipcm_cookie { 58 __be32 addr; 59 int oif; 60 struct ip_options_rcu *opt; 61 __u8 tx_flags; 62 __u8 ttl; 63 __s16 tos; 64 char priority; 65}; 66 67#define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) 68#define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) 69 70struct ip_ra_chain { 71 struct ip_ra_chain __rcu *next; 72 struct sock *sk; 73 union { 74 void (*destructor)(struct sock *); 75 struct sock *saved_sk; 76 }; 77 struct rcu_head rcu; 78}; 79 80extern struct ip_ra_chain __rcu *ip_ra_chain; 81 82/* IP flags. */ 83#define IP_CE 0x8000 /* Flag: "Congestion" */ 84#define IP_DF 0x4000 /* Flag: "Don't Fragment" */ 85#define IP_MF 0x2000 /* Flag: "More Fragments" */ 86#define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ 87 88#define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ 89 90struct msghdr; 91struct net_device; 92struct packet_type; 93struct rtable; 94struct sockaddr; 95 96int igmp_mc_init(void); 97 98/* 99 * Functions provided by ip.c 100 */ 101 102int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk, 103 __be32 saddr, __be32 daddr, 104 struct ip_options_rcu *opt); 105int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, 106 struct net_device *orig_dev); 107int ip_local_deliver(struct sk_buff *skb); 108int ip_mr_input(struct sk_buff *skb); 109int ip_output(struct sock *sk, struct sk_buff *skb); 110int ip_mc_output(struct sock *sk, struct sk_buff *skb); 111int ip_fragment(struct sock *sk, struct sk_buff *skb, 112 int (*output)(struct sock *, struct sk_buff *)); 113int ip_do_nat(struct sk_buff *skb); 114void ip_send_check(struct iphdr *ip); 115int __ip_local_out(struct sk_buff *skb); 116int ip_local_out_sk(struct sock *sk, struct sk_buff *skb); 117static inline int ip_local_out(struct sk_buff *skb) 118{ 119 return ip_local_out_sk(skb->sk, skb); 120} 121 122int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); 123void ip_init(void); 124int ip_append_data(struct sock *sk, struct flowi4 *fl4, 125 int getfrag(void *from, char *to, int offset, int len, 126 int odd, struct sk_buff *skb), 127 void *from, int len, int protolen, 128 struct ipcm_cookie *ipc, 129 struct rtable **rt, 130 unsigned int flags); 131int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, 132 struct sk_buff *skb); 133ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page, 134 int offset, size_t size, int flags); 135struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, 136 struct sk_buff_head *queue, 137 struct inet_cork *cork); 138int ip_send_skb(struct net *net, struct sk_buff *skb); 139int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); 140void ip_flush_pending_frames(struct sock *sk); 141struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, 142 int getfrag(void *from, char *to, int offset, 143 int len, int odd, struct sk_buff *skb), 144 void *from, int length, int transhdrlen, 145 struct ipcm_cookie *ipc, struct rtable **rtp, 146 unsigned int flags); 147 148static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) 149{ 150 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base); 151} 152 153static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet) 154{ 155 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos); 156} 157 158static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk) 159{ 160 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk); 161} 162 163/* datagram.c */ 164int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 165int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 166 167void ip4_datagram_release_cb(struct sock *sk); 168 169struct ip_reply_arg { 170 struct kvec iov[1]; 171 int flags; 172 __wsum csum; 173 int csumoffset; /* u16 offset of csum in iov[0].iov_base */ 174 /* -1 if not needed */ 175 int bound_dev_if; 176 u8 tos; 177}; 178 179#define IP_REPLY_ARG_NOSRCCHECK 1 180 181static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg) 182{ 183 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0; 184} 185 186void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, 187 const struct ip_options *sopt, 188 __be32 daddr, __be32 saddr, 189 const struct ip_reply_arg *arg, 190 unsigned int len); 191 192#define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) 193#define IP_INC_STATS_BH(net, field) SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field) 194#define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 195#define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val) 196#define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 197#define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val) 198#define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) 199#define NET_INC_STATS_BH(net, field) SNMP_INC_STATS_BH((net)->mib.net_statistics, field) 200#define NET_INC_STATS_USER(net, field) SNMP_INC_STATS_USER((net)->mib.net_statistics, field) 201#define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 202#define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd) 203#define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd) 204 205unsigned long snmp_fold_field(void __percpu *mib, int offt); 206#if BITS_PER_LONG==32 207u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); 208#else 209static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) 210{ 211 return snmp_fold_field(mib, offt); 212} 213#endif 214 215void inet_get_local_port_range(struct net *net, int *low, int *high); 216 217#ifdef CONFIG_SYSCTL 218static inline int inet_is_local_reserved_port(struct net *net, int port) 219{ 220 if (!net->ipv4.sysctl_local_reserved_ports) 221 return 0; 222 return test_bit(port, net->ipv4.sysctl_local_reserved_ports); 223} 224 225static inline bool sysctl_dev_name_is_allowed(const char *name) 226{ 227 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; 228} 229 230#else 231static inline int inet_is_local_reserved_port(struct net *net, int port) 232{ 233 return 0; 234} 235#endif 236 237/* From inetpeer.c */ 238extern int inet_peer_threshold; 239extern int inet_peer_minttl; 240extern int inet_peer_maxttl; 241 242/* From ip_input.c */ 243extern int sysctl_ip_early_demux; 244 245/* From ip_output.c */ 246extern int sysctl_ip_dynaddr; 247 248void ipfrag_init(void); 249 250void ip_static_sysctl_init(void); 251 252#define IP4_REPLY_MARK(net, mark) \ 253 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0) 254 255static inline bool ip_is_fragment(const struct iphdr *iph) 256{ 257 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; 258} 259 260#ifdef CONFIG_INET 261#include <net/dst.h> 262 263/* The function in 2.2 was invalid, producing wrong result for 264 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ 265static inline 266int ip_decrease_ttl(struct iphdr *iph) 267{ 268 u32 check = (__force u32)iph->check; 269 check += (__force u32)htons(0x0100); 270 iph->check = (__force __sum16)(check + (check>=0xFFFF)); 271 return --iph->ttl; 272} 273 274static inline 275int ip_dont_fragment(struct sock *sk, struct dst_entry *dst) 276{ 277 return inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO || 278 (inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT && 279 !(dst_metric_locked(dst, RTAX_MTU))); 280} 281 282static inline bool ip_sk_accept_pmtu(const struct sock *sk) 283{ 284 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE && 285 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT; 286} 287 288static inline bool ip_sk_use_pmtu(const struct sock *sk) 289{ 290 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE; 291} 292 293static inline bool ip_sk_ignore_df(const struct sock *sk) 294{ 295 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO || 296 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT; 297} 298 299static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, 300 bool forwarding) 301{ 302 struct net *net = dev_net(dst->dev); 303 304 if (net->ipv4.sysctl_ip_fwd_use_pmtu || 305 dst_metric_locked(dst, RTAX_MTU) || 306 !forwarding) 307 return dst_mtu(dst); 308 309 return min(dst->dev->mtu, IP_MAX_MTU); 310} 311 312static inline unsigned int ip_skb_dst_mtu(const struct sk_buff *skb) 313{ 314 if (!skb->sk || ip_sk_use_pmtu(skb->sk)) { 315 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; 316 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding); 317 } else { 318 return min(skb_dst(skb)->dev->mtu, IP_MAX_MTU); 319 } 320} 321 322u32 ip_idents_reserve(u32 hash, int segs); 323void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); 324 325static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, 326 struct sock *sk, int segs) 327{ 328 struct iphdr *iph = ip_hdr(skb); 329 330 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { 331 /* This is only to work around buggy Windows95/2000 332 * VJ compression implementations. If the ID field 333 * does not change, they drop every other packet in 334 * a TCP stream using header compression. 335 */ 336 if (sk && inet_sk(sk)->inet_daddr) { 337 iph->id = htons(inet_sk(sk)->inet_id); 338 inet_sk(sk)->inet_id += segs; 339 } else { 340 iph->id = 0; 341 } 342 } else { 343 __ip_select_ident(net, iph, segs); 344 } 345} 346 347static inline void ip_select_ident(struct net *net, struct sk_buff *skb, 348 struct sock *sk) 349{ 350 ip_select_ident_segs(net, skb, sk, 1); 351} 352 353static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) 354{ 355 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, 356 skb->len, proto, 0); 357} 358 359static inline void inet_set_txhash(struct sock *sk) 360{ 361 struct inet_sock *inet = inet_sk(sk); 362 struct flow_keys keys; 363 364 keys.src = inet->inet_saddr; 365 keys.dst = inet->inet_daddr; 366 keys.port16[0] = inet->inet_sport; 367 keys.port16[1] = inet->inet_dport; 368 369 sk->sk_txhash = flow_hash_from_keys(&keys); 370} 371 372static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto) 373{ 374 const struct iphdr *iph = skb_gro_network_header(skb); 375 376 return csum_tcpudp_nofold(iph->saddr, iph->daddr, 377 skb_gro_len(skb), proto, 0); 378} 379 380/* 381 * Map a multicast IP onto multicast MAC for type ethernet. 382 */ 383 384static inline void ip_eth_mc_map(__be32 naddr, char *buf) 385{ 386 __u32 addr=ntohl(naddr); 387 buf[0]=0x01; 388 buf[1]=0x00; 389 buf[2]=0x5e; 390 buf[5]=addr&0xFF; 391 addr>>=8; 392 buf[4]=addr&0xFF; 393 addr>>=8; 394 buf[3]=addr&0x7F; 395} 396 397/* 398 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. 399 * Leave P_Key as 0 to be filled in by driver. 400 */ 401 402static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 403{ 404 __u32 addr; 405 unsigned char scope = broadcast[5] & 0xF; 406 407 buf[0] = 0; /* Reserved */ 408 buf[1] = 0xff; /* Multicast QPN */ 409 buf[2] = 0xff; 410 buf[3] = 0xff; 411 addr = ntohl(naddr); 412 buf[4] = 0xff; 413 buf[5] = 0x10 | scope; /* scope from broadcast address */ 414 buf[6] = 0x40; /* IPv4 signature */ 415 buf[7] = 0x1b; 416 buf[8] = broadcast[8]; /* P_Key */ 417 buf[9] = broadcast[9]; 418 buf[10] = 0; 419 buf[11] = 0; 420 buf[12] = 0; 421 buf[13] = 0; 422 buf[14] = 0; 423 buf[15] = 0; 424 buf[19] = addr & 0xff; 425 addr >>= 8; 426 buf[18] = addr & 0xff; 427 addr >>= 8; 428 buf[17] = addr & 0xff; 429 addr >>= 8; 430 buf[16] = addr & 0x0f; 431} 432 433static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 434{ 435 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) 436 memcpy(buf, broadcast, 4); 437 else 438 memcpy(buf, &naddr, sizeof(naddr)); 439} 440 441#if IS_ENABLED(CONFIG_IPV6) 442#include <linux/ipv6.h> 443#endif 444 445static __inline__ void inet_reset_saddr(struct sock *sk) 446{ 447 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; 448#if IS_ENABLED(CONFIG_IPV6) 449 if (sk->sk_family == PF_INET6) { 450 struct ipv6_pinfo *np = inet6_sk(sk); 451 452 memset(&np->saddr, 0, sizeof(np->saddr)); 453 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); 454 } 455#endif 456} 457 458#endif 459 460bool ip_call_ra_chain(struct sk_buff *skb); 461 462/* 463 * Functions provided by ip_fragment.c 464 */ 465 466enum ip_defrag_users { 467 IP_DEFRAG_LOCAL_DELIVER, 468 IP_DEFRAG_CALL_RA_CHAIN, 469 IP_DEFRAG_CONNTRACK_IN, 470 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, 471 IP_DEFRAG_CONNTRACK_OUT, 472 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 473 IP_DEFRAG_CONNTRACK_BRIDGE_IN, 474 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 475 IP_DEFRAG_VS_IN, 476 IP_DEFRAG_VS_OUT, 477 IP_DEFRAG_VS_FWD, 478 IP_DEFRAG_AF_PACKET, 479 IP_DEFRAG_MACVLAN, 480}; 481 482int ip_defrag(struct sk_buff *skb, u32 user); 483#ifdef CONFIG_INET 484struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user); 485#else 486static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user) 487{ 488 return skb; 489} 490#endif 491int ip_frag_mem(struct net *net); 492 493/* 494 * Functions provided by ip_forward.c 495 */ 496 497int ip_forward(struct sk_buff *skb); 498 499/* 500 * Functions provided by ip_options.c 501 */ 502 503void ip_options_build(struct sk_buff *skb, struct ip_options *opt, 504 __be32 daddr, struct rtable *rt, int is_frag); 505 506int __ip_options_echo(struct ip_options *dopt, struct sk_buff *skb, 507 const struct ip_options *sopt); 508static inline int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb) 509{ 510 return __ip_options_echo(dopt, skb, &IPCB(skb)->opt); 511} 512 513void ip_options_fragment(struct sk_buff *skb); 514int ip_options_compile(struct net *net, struct ip_options *opt, 515 struct sk_buff *skb); 516int ip_options_get(struct net *net, struct ip_options_rcu **optp, 517 unsigned char *data, int optlen); 518int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp, 519 unsigned char __user *data, int optlen); 520void ip_options_undo(struct ip_options *opt); 521void ip_forward_options(struct sk_buff *skb); 522int ip_options_rcv_srr(struct sk_buff *skb); 523 524/* 525 * Functions provided by ip_sockglue.c 526 */ 527 528void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb); 529void ip_cmsg_recv_offset(struct msghdr *msg, struct sk_buff *skb, int offset); 530int ip_cmsg_send(struct net *net, struct msghdr *msg, 531 struct ipcm_cookie *ipc, bool allow_ipv6); 532int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, 533 unsigned int optlen); 534int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, 535 int __user *optlen); 536int compat_ip_setsockopt(struct sock *sk, int level, int optname, 537 char __user *optval, unsigned int optlen); 538int compat_ip_getsockopt(struct sock *sk, int level, int optname, 539 char __user *optval, int __user *optlen); 540int ip_ra_control(struct sock *sk, unsigned char on, 541 void (*destructor)(struct sock *)); 542 543int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); 544void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 545 u32 info, u8 *payload); 546void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, 547 u32 info); 548 549static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 550{ 551 ip_cmsg_recv_offset(msg, skb, 0); 552} 553 554bool icmp_global_allow(void); 555extern int sysctl_icmp_msgs_per_sec; 556extern int sysctl_icmp_msgs_burst; 557 558#ifdef CONFIG_PROC_FS 559int ip_misc_proc_init(void); 560#endif 561 562#endif /* _IP_H */ 563