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 int iif; 40 struct ip_options opt; /* Compiled IP options */ 41 unsigned char flags; 42 43 #define IPSKB_FORWARDED BIT(0) 44 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1) 45 #define IPSKB_XFRM_TRANSFORMED BIT(2) 46 #define IPSKB_FRAG_COMPLETE BIT(3) 47 #define IPSKB_REROUTED BIT(4) 48 #define IPSKB_DOREDIRECT BIT(5) 49 #define IPSKB_FRAG_PMTU BIT(6) 50 51 u16 frag_max_size; 52 }; 53 54 static inline unsigned int ip_hdrlen(const struct sk_buff *skb) 55 { 56 return ip_hdr(skb)->ihl * 4; 57 } 58 59 struct ipcm_cookie { 60 struct sockcm_cookie sockc; 61 __be32 addr; 62 int oif; 63 struct ip_options_rcu *opt; 64 __u8 tx_flags; 65 __u8 ttl; 66 __s16 tos; 67 char priority; 68 }; 69 70 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) 71 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) 72 73 struct ip_ra_chain { 74 struct ip_ra_chain __rcu *next; 75 struct sock *sk; 76 union { 77 void (*destructor)(struct sock *); 78 struct sock *saved_sk; 79 }; 80 struct rcu_head rcu; 81 }; 82 83 extern struct ip_ra_chain __rcu *ip_ra_chain; 84 85 /* IP flags. */ 86 #define IP_CE 0x8000 /* Flag: "Congestion" */ 87 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */ 88 #define IP_MF 0x2000 /* Flag: "More Fragments" */ 89 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ 90 91 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ 92 93 struct msghdr; 94 struct net_device; 95 struct packet_type; 96 struct rtable; 97 struct sockaddr; 98 99 int igmp_mc_init(void); 100 101 /* 102 * Functions provided by ip.c 103 */ 104 105 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk, 106 __be32 saddr, __be32 daddr, 107 struct ip_options_rcu *opt); 108 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, 109 struct net_device *orig_dev); 110 int ip_local_deliver(struct sk_buff *skb); 111 int ip_mr_input(struct sk_buff *skb); 112 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb); 113 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb); 114 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, 115 int (*output)(struct net *, struct sock *, struct sk_buff *)); 116 void ip_send_check(struct iphdr *ip); 117 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 118 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 119 120 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); 121 void ip_init(void); 122 int ip_append_data(struct sock *sk, struct flowi4 *fl4, 123 int getfrag(void *from, char *to, int offset, int len, 124 int odd, struct sk_buff *skb), 125 void *from, int len, int protolen, 126 struct ipcm_cookie *ipc, 127 struct rtable **rt, 128 unsigned int flags); 129 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, 130 struct sk_buff *skb); 131 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page, 132 int offset, size_t size, int flags); 133 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, 134 struct sk_buff_head *queue, 135 struct inet_cork *cork); 136 int ip_send_skb(struct net *net, struct sk_buff *skb); 137 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); 138 void ip_flush_pending_frames(struct sock *sk); 139 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, 140 int getfrag(void *from, char *to, int offset, 141 int len, int odd, struct sk_buff *skb), 142 void *from, int length, int transhdrlen, 143 struct ipcm_cookie *ipc, struct rtable **rtp, 144 unsigned int flags); 145 146 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) 147 { 148 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base); 149 } 150 151 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet) 152 { 153 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos); 154 } 155 156 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk) 157 { 158 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk); 159 } 160 161 /* datagram.c */ 162 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 163 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 164 165 void ip4_datagram_release_cb(struct sock *sk); 166 167 struct ip_reply_arg { 168 struct kvec iov[1]; 169 int flags; 170 __wsum csum; 171 int csumoffset; /* u16 offset of csum in iov[0].iov_base */ 172 /* -1 if not needed */ 173 int bound_dev_if; 174 u8 tos; 175 }; 176 177 #define IP_REPLY_ARG_NOSRCCHECK 1 178 179 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg) 180 { 181 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0; 182 } 183 184 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, 185 const struct ip_options *sopt, 186 __be32 daddr, __be32 saddr, 187 const struct ip_reply_arg *arg, 188 unsigned int len); 189 190 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) 191 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field) 192 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 193 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 194 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 195 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 196 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) 197 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field) 198 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 199 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((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(struct sock *sk, 317 const struct sk_buff *skb) 318 { 319 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) { 320 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; 321 322 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding); 323 } 324 325 return min(skb_dst(skb)->dev->mtu, IP_MAX_MTU); 326 } 327 328 u32 ip_idents_reserve(u32 hash, int segs); 329 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); 330 331 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, 332 struct sock *sk, int segs) 333 { 334 struct iphdr *iph = ip_hdr(skb); 335 336 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { 337 /* This is only to work around buggy Windows95/2000 338 * VJ compression implementations. If the ID field 339 * does not change, they drop every other packet in 340 * a TCP stream using header compression. 341 */ 342 if (sk && inet_sk(sk)->inet_daddr) { 343 iph->id = htons(inet_sk(sk)->inet_id); 344 inet_sk(sk)->inet_id += segs; 345 } else { 346 iph->id = 0; 347 } 348 } else { 349 __ip_select_ident(net, iph, segs); 350 } 351 } 352 353 static inline void ip_select_ident(struct net *net, struct sk_buff *skb, 354 struct sock *sk) 355 { 356 ip_select_ident_segs(net, skb, sk, 1); 357 } 358 359 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) 360 { 361 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, 362 skb->len, proto, 0); 363 } 364 365 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store 366 * Equivalent to : flow->v4addrs.src = iph->saddr; 367 * flow->v4addrs.dst = iph->daddr; 368 */ 369 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, 370 const struct iphdr *iph) 371 { 372 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != 373 offsetof(typeof(flow->addrs), v4addrs.src) + 374 sizeof(flow->addrs.v4addrs.src)); 375 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs)); 376 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; 377 } 378 379 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto) 380 { 381 const struct iphdr *iph = skb_gro_network_header(skb); 382 383 return csum_tcpudp_nofold(iph->saddr, iph->daddr, 384 skb_gro_len(skb), proto, 0); 385 } 386 387 /* 388 * Map a multicast IP onto multicast MAC for type ethernet. 389 */ 390 391 static inline void ip_eth_mc_map(__be32 naddr, char *buf) 392 { 393 __u32 addr=ntohl(naddr); 394 buf[0]=0x01; 395 buf[1]=0x00; 396 buf[2]=0x5e; 397 buf[5]=addr&0xFF; 398 addr>>=8; 399 buf[4]=addr&0xFF; 400 addr>>=8; 401 buf[3]=addr&0x7F; 402 } 403 404 /* 405 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. 406 * Leave P_Key as 0 to be filled in by driver. 407 */ 408 409 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 410 { 411 __u32 addr; 412 unsigned char scope = broadcast[5] & 0xF; 413 414 buf[0] = 0; /* Reserved */ 415 buf[1] = 0xff; /* Multicast QPN */ 416 buf[2] = 0xff; 417 buf[3] = 0xff; 418 addr = ntohl(naddr); 419 buf[4] = 0xff; 420 buf[5] = 0x10 | scope; /* scope from broadcast address */ 421 buf[6] = 0x40; /* IPv4 signature */ 422 buf[7] = 0x1b; 423 buf[8] = broadcast[8]; /* P_Key */ 424 buf[9] = broadcast[9]; 425 buf[10] = 0; 426 buf[11] = 0; 427 buf[12] = 0; 428 buf[13] = 0; 429 buf[14] = 0; 430 buf[15] = 0; 431 buf[19] = addr & 0xff; 432 addr >>= 8; 433 buf[18] = addr & 0xff; 434 addr >>= 8; 435 buf[17] = addr & 0xff; 436 addr >>= 8; 437 buf[16] = addr & 0x0f; 438 } 439 440 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 441 { 442 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) 443 memcpy(buf, broadcast, 4); 444 else 445 memcpy(buf, &naddr, sizeof(naddr)); 446 } 447 448 #if IS_ENABLED(CONFIG_IPV6) 449 #include <linux/ipv6.h> 450 #endif 451 452 static __inline__ void inet_reset_saddr(struct sock *sk) 453 { 454 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; 455 #if IS_ENABLED(CONFIG_IPV6) 456 if (sk->sk_family == PF_INET6) { 457 struct ipv6_pinfo *np = inet6_sk(sk); 458 459 memset(&np->saddr, 0, sizeof(np->saddr)); 460 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); 461 } 462 #endif 463 } 464 465 #endif 466 467 static inline unsigned int ipv4_addr_hash(__be32 ip) 468 { 469 return (__force unsigned int) ip; 470 } 471 472 bool ip_call_ra_chain(struct sk_buff *skb); 473 474 /* 475 * Functions provided by ip_fragment.c 476 */ 477 478 enum ip_defrag_users { 479 IP_DEFRAG_LOCAL_DELIVER, 480 IP_DEFRAG_CALL_RA_CHAIN, 481 IP_DEFRAG_CONNTRACK_IN, 482 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, 483 IP_DEFRAG_CONNTRACK_OUT, 484 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 485 IP_DEFRAG_CONNTRACK_BRIDGE_IN, 486 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 487 IP_DEFRAG_VS_IN, 488 IP_DEFRAG_VS_OUT, 489 IP_DEFRAG_VS_FWD, 490 IP_DEFRAG_AF_PACKET, 491 IP_DEFRAG_MACVLAN, 492 }; 493 494 /* Return true if the value of 'user' is between 'lower_bond' 495 * and 'upper_bond' inclusively. 496 */ 497 static inline bool ip_defrag_user_in_between(u32 user, 498 enum ip_defrag_users lower_bond, 499 enum ip_defrag_users upper_bond) 500 { 501 return user >= lower_bond && user <= upper_bond; 502 } 503 504 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user); 505 #ifdef CONFIG_INET 506 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user); 507 #else 508 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user) 509 { 510 return skb; 511 } 512 #endif 513 int ip_frag_mem(struct net *net); 514 515 /* 516 * Functions provided by ip_forward.c 517 */ 518 519 int ip_forward(struct sk_buff *skb); 520 521 /* 522 * Functions provided by ip_options.c 523 */ 524 525 void ip_options_build(struct sk_buff *skb, struct ip_options *opt, 526 __be32 daddr, struct rtable *rt, int is_frag); 527 528 int __ip_options_echo(struct ip_options *dopt, struct sk_buff *skb, 529 const struct ip_options *sopt); 530 static inline int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb) 531 { 532 return __ip_options_echo(dopt, skb, &IPCB(skb)->opt); 533 } 534 535 void ip_options_fragment(struct sk_buff *skb); 536 int ip_options_compile(struct net *net, struct ip_options *opt, 537 struct sk_buff *skb); 538 int ip_options_get(struct net *net, struct ip_options_rcu **optp, 539 unsigned char *data, int optlen); 540 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp, 541 unsigned char __user *data, int optlen); 542 void ip_options_undo(struct ip_options *opt); 543 void ip_forward_options(struct sk_buff *skb); 544 int ip_options_rcv_srr(struct sk_buff *skb); 545 546 /* 547 * Functions provided by ip_sockglue.c 548 */ 549 550 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb); 551 void ip_cmsg_recv_offset(struct msghdr *msg, struct sk_buff *skb, int offset); 552 int ip_cmsg_send(struct sock *sk, struct msghdr *msg, 553 struct ipcm_cookie *ipc, bool allow_ipv6); 554 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, 555 unsigned int optlen); 556 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, 557 int __user *optlen); 558 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 559 char __user *optval, unsigned int optlen); 560 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 561 char __user *optval, int __user *optlen); 562 int ip_ra_control(struct sock *sk, unsigned char on, 563 void (*destructor)(struct sock *)); 564 565 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); 566 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 567 u32 info, u8 *payload); 568 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, 569 u32 info); 570 571 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 572 { 573 ip_cmsg_recv_offset(msg, skb, 0); 574 } 575 576 bool icmp_global_allow(void); 577 extern int sysctl_icmp_msgs_per_sec; 578 extern int sysctl_icmp_msgs_burst; 579 580 #ifdef CONFIG_PROC_FS 581 int ip_misc_proc_init(void); 582 #endif 583 584 #endif /* _IP_H */ 585