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, 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 sock *sk, struct sk_buff *skb); 111 int ip_mc_output(struct sock *sk, struct sk_buff *skb); 112 int ip_do_fragment(struct sock *sk, struct sk_buff *skb, 113 int (*output)(struct sock *, struct sk_buff *)); 114 void ip_send_check(struct iphdr *ip); 115 int __ip_local_out(struct sk_buff *skb); 116 int ip_local_out_sk(struct sock *sk, struct sk_buff *skb); 117 static inline int ip_local_out(struct sk_buff *skb) 118 { 119 return ip_local_out_sk(skb->sk, skb); 120 } 121 122 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); 123 void ip_init(void); 124 int 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); 131 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, 132 struct sk_buff *skb); 133 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page, 134 int offset, size_t size, int flags); 135 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, 136 struct sk_buff_head *queue, 137 struct inet_cork *cork); 138 int ip_send_skb(struct net *net, struct sk_buff *skb); 139 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); 140 void ip_flush_pending_frames(struct sock *sk); 141 struct 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); 147 148 static 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 } 152 153 static 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 } 157 158 static 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 */ 164 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 165 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 166 167 void ip4_datagram_release_cb(struct sock *sk); 168 169 struct 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 1 180 181 static 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 } 185 186 void 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) 204 205 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct); 206 unsigned long snmp_fold_field(void __percpu *mib, int offt); 207 #if BITS_PER_LONG==32 208 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 209 size_t syncp_offset); 210 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); 211 #else 212 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 213 size_t syncp_offset) 214 { 215 return snmp_get_cpu_field(mib, cpu, offct); 216 217 } 218 219 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) 220 { 221 return snmp_fold_field(mib, offt); 222 } 223 #endif 224 225 void inet_get_local_port_range(struct net *net, int *low, int *high); 226 227 #ifdef CONFIG_SYSCTL 228 static inline int inet_is_local_reserved_port(struct net *net, int port) 229 { 230 if (!net->ipv4.sysctl_local_reserved_ports) 231 return 0; 232 return test_bit(port, net->ipv4.sysctl_local_reserved_ports); 233 } 234 235 static inline bool sysctl_dev_name_is_allowed(const char *name) 236 { 237 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; 238 } 239 240 #else 241 static inline int inet_is_local_reserved_port(struct net *net, int port) 242 { 243 return 0; 244 } 245 #endif 246 247 /* From inetpeer.c */ 248 extern int inet_peer_threshold; 249 extern int inet_peer_minttl; 250 extern int inet_peer_maxttl; 251 252 /* From ip_input.c */ 253 extern int sysctl_ip_early_demux; 254 255 /* From ip_output.c */ 256 extern int sysctl_ip_dynaddr; 257 258 void ipfrag_init(void); 259 260 void ip_static_sysctl_init(void); 261 262 #define IP4_REPLY_MARK(net, mark) \ 263 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0) 264 265 static inline bool ip_is_fragment(const struct iphdr *iph) 266 { 267 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; 268 } 269 270 #ifdef CONFIG_INET 271 #include <net/dst.h> 272 273 /* The function in 2.2 was invalid, producing wrong result for 274 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ 275 static inline 276 int ip_decrease_ttl(struct iphdr *iph) 277 { 278 u32 check = (__force u32)iph->check; 279 check += (__force u32)htons(0x0100); 280 iph->check = (__force __sum16)(check + (check>=0xFFFF)); 281 return --iph->ttl; 282 } 283 284 static inline 285 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst) 286 { 287 return inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO || 288 (inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT && 289 !(dst_metric_locked(dst, RTAX_MTU))); 290 } 291 292 static inline bool ip_sk_accept_pmtu(const struct sock *sk) 293 { 294 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE && 295 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT; 296 } 297 298 static inline bool ip_sk_use_pmtu(const struct sock *sk) 299 { 300 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE; 301 } 302 303 static inline bool ip_sk_ignore_df(const struct sock *sk) 304 { 305 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO || 306 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT; 307 } 308 309 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, 310 bool forwarding) 311 { 312 struct net *net = dev_net(dst->dev); 313 314 if (net->ipv4.sysctl_ip_fwd_use_pmtu || 315 dst_metric_locked(dst, RTAX_MTU) || 316 !forwarding) 317 return dst_mtu(dst); 318 319 return min(dst->dev->mtu, IP_MAX_MTU); 320 } 321 322 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff *skb) 323 { 324 if (!skb->sk || ip_sk_use_pmtu(skb->sk)) { 325 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; 326 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding); 327 } else { 328 return min(skb_dst(skb)->dev->mtu, IP_MAX_MTU); 329 } 330 } 331 332 u32 ip_idents_reserve(u32 hash, int segs); 333 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); 334 335 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, 336 struct sock *sk, int segs) 337 { 338 struct iphdr *iph = ip_hdr(skb); 339 340 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { 341 /* This is only to work around buggy Windows95/2000 342 * VJ compression implementations. If the ID field 343 * does not change, they drop every other packet in 344 * a TCP stream using header compression. 345 */ 346 if (sk && inet_sk(sk)->inet_daddr) { 347 iph->id = htons(inet_sk(sk)->inet_id); 348 inet_sk(sk)->inet_id += segs; 349 } else { 350 iph->id = 0; 351 } 352 } else { 353 __ip_select_ident(net, iph, segs); 354 } 355 } 356 357 static inline void ip_select_ident(struct net *net, struct sk_buff *skb, 358 struct sock *sk) 359 { 360 ip_select_ident_segs(net, skb, sk, 1); 361 } 362 363 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) 364 { 365 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, 366 skb->len, proto, 0); 367 } 368 369 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store 370 * Equivalent to : flow->v4addrs.src = iph->saddr; 371 * flow->v4addrs.dst = iph->daddr; 372 */ 373 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, 374 const struct iphdr *iph) 375 { 376 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != 377 offsetof(typeof(flow->addrs), v4addrs.src) + 378 sizeof(flow->addrs.v4addrs.src)); 379 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs)); 380 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; 381 } 382 383 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto) 384 { 385 const struct iphdr *iph = skb_gro_network_header(skb); 386 387 return csum_tcpudp_nofold(iph->saddr, iph->daddr, 388 skb_gro_len(skb), proto, 0); 389 } 390 391 /* 392 * Map a multicast IP onto multicast MAC for type ethernet. 393 */ 394 395 static inline void ip_eth_mc_map(__be32 naddr, char *buf) 396 { 397 __u32 addr=ntohl(naddr); 398 buf[0]=0x01; 399 buf[1]=0x00; 400 buf[2]=0x5e; 401 buf[5]=addr&0xFF; 402 addr>>=8; 403 buf[4]=addr&0xFF; 404 addr>>=8; 405 buf[3]=addr&0x7F; 406 } 407 408 /* 409 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. 410 * Leave P_Key as 0 to be filled in by driver. 411 */ 412 413 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 414 { 415 __u32 addr; 416 unsigned char scope = broadcast[5] & 0xF; 417 418 buf[0] = 0; /* Reserved */ 419 buf[1] = 0xff; /* Multicast QPN */ 420 buf[2] = 0xff; 421 buf[3] = 0xff; 422 addr = ntohl(naddr); 423 buf[4] = 0xff; 424 buf[5] = 0x10 | scope; /* scope from broadcast address */ 425 buf[6] = 0x40; /* IPv4 signature */ 426 buf[7] = 0x1b; 427 buf[8] = broadcast[8]; /* P_Key */ 428 buf[9] = broadcast[9]; 429 buf[10] = 0; 430 buf[11] = 0; 431 buf[12] = 0; 432 buf[13] = 0; 433 buf[14] = 0; 434 buf[15] = 0; 435 buf[19] = addr & 0xff; 436 addr >>= 8; 437 buf[18] = addr & 0xff; 438 addr >>= 8; 439 buf[17] = addr & 0xff; 440 addr >>= 8; 441 buf[16] = addr & 0x0f; 442 } 443 444 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 445 { 446 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) 447 memcpy(buf, broadcast, 4); 448 else 449 memcpy(buf, &naddr, sizeof(naddr)); 450 } 451 452 #if IS_ENABLED(CONFIG_IPV6) 453 #include <linux/ipv6.h> 454 #endif 455 456 static __inline__ void inet_reset_saddr(struct sock *sk) 457 { 458 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; 459 #if IS_ENABLED(CONFIG_IPV6) 460 if (sk->sk_family == PF_INET6) { 461 struct ipv6_pinfo *np = inet6_sk(sk); 462 463 memset(&np->saddr, 0, sizeof(np->saddr)); 464 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); 465 } 466 #endif 467 } 468 469 #endif 470 471 static inline unsigned int ipv4_addr_hash(__be32 ip) 472 { 473 return (__force unsigned int) ip; 474 } 475 476 bool ip_call_ra_chain(struct sk_buff *skb); 477 478 /* 479 * Functions provided by ip_fragment.c 480 */ 481 482 enum ip_defrag_users { 483 IP_DEFRAG_LOCAL_DELIVER, 484 IP_DEFRAG_CALL_RA_CHAIN, 485 IP_DEFRAG_CONNTRACK_IN, 486 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, 487 IP_DEFRAG_CONNTRACK_OUT, 488 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 489 IP_DEFRAG_CONNTRACK_BRIDGE_IN, 490 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 491 IP_DEFRAG_VS_IN, 492 IP_DEFRAG_VS_OUT, 493 IP_DEFRAG_VS_FWD, 494 IP_DEFRAG_AF_PACKET, 495 IP_DEFRAG_MACVLAN, 496 }; 497 498 /* Return true if the value of 'user' is between 'lower_bond' 499 * and 'upper_bond' inclusively. 500 */ 501 static inline bool ip_defrag_user_in_between(u32 user, 502 enum ip_defrag_users lower_bond, 503 enum ip_defrag_users upper_bond) 504 { 505 return user >= lower_bond && user <= upper_bond; 506 } 507 508 int ip_defrag(struct sk_buff *skb, u32 user); 509 #ifdef CONFIG_INET 510 struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user); 511 #else 512 static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user) 513 { 514 return skb; 515 } 516 #endif 517 int ip_frag_mem(struct net *net); 518 519 /* 520 * Functions provided by ip_forward.c 521 */ 522 523 int ip_forward(struct sk_buff *skb); 524 525 /* 526 * Functions provided by ip_options.c 527 */ 528 529 void ip_options_build(struct sk_buff *skb, struct ip_options *opt, 530 __be32 daddr, struct rtable *rt, int is_frag); 531 532 int __ip_options_echo(struct ip_options *dopt, struct sk_buff *skb, 533 const struct ip_options *sopt); 534 static inline int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb) 535 { 536 return __ip_options_echo(dopt, skb, &IPCB(skb)->opt); 537 } 538 539 void ip_options_fragment(struct sk_buff *skb); 540 int ip_options_compile(struct net *net, struct ip_options *opt, 541 struct sk_buff *skb); 542 int ip_options_get(struct net *net, struct ip_options_rcu **optp, 543 unsigned char *data, int optlen); 544 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp, 545 unsigned char __user *data, int optlen); 546 void ip_options_undo(struct ip_options *opt); 547 void ip_forward_options(struct sk_buff *skb); 548 int ip_options_rcv_srr(struct sk_buff *skb); 549 550 /* 551 * Functions provided by ip_sockglue.c 552 */ 553 554 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb); 555 void ip_cmsg_recv_offset(struct msghdr *msg, struct sk_buff *skb, int offset); 556 int ip_cmsg_send(struct net *net, struct msghdr *msg, 557 struct ipcm_cookie *ipc, bool allow_ipv6); 558 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, 559 unsigned int optlen); 560 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, 561 int __user *optlen); 562 int compat_ip_setsockopt(struct sock *sk, int level, int optname, 563 char __user *optval, unsigned int optlen); 564 int compat_ip_getsockopt(struct sock *sk, int level, int optname, 565 char __user *optval, int __user *optlen); 566 int ip_ra_control(struct sock *sk, unsigned char on, 567 void (*destructor)(struct sock *)); 568 569 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); 570 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 571 u32 info, u8 *payload); 572 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, 573 u32 info); 574 575 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 576 { 577 ip_cmsg_recv_offset(msg, skb, 0); 578 } 579 580 bool icmp_global_allow(void); 581 extern int sysctl_icmp_msgs_per_sec; 582 extern int sysctl_icmp_msgs_burst; 583 584 #ifdef CONFIG_PROC_FS 585 int ip_misc_proc_init(void); 586 #endif 587 588 #endif /* _IP_H */ 589