1 /* SPDX-License-Identifier: GPL-2.0-or-later */ 2 /* 3 * INET An implementation of the TCP/IP protocol suite for the LINUX 4 * operating system. INET is implemented using the BSD Socket 5 * interface as the means of communication with the user level. 6 * 7 * Definitions for the IP module. 8 * 9 * Version: @(#)ip.h 1.0.2 05/07/93 10 * 11 * Authors: Ross Biro 12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> 13 * Alan Cox, <gw4pts@gw4pts.ampr.org> 14 * 15 * Changes: 16 * Mike McLagan : Routing by source 17 */ 18 #ifndef _IP_H 19 #define _IP_H 20 21 #include <linux/types.h> 22 #include <linux/ip.h> 23 #include <linux/in.h> 24 #include <linux/skbuff.h> 25 #include <linux/jhash.h> 26 #include <linux/sockptr.h> 27 #include <linux/static_key.h> 28 29 #include <net/inet_sock.h> 30 #include <net/route.h> 31 #include <net/snmp.h> 32 #include <net/flow.h> 33 #include <net/flow_dissector.h> 34 #include <net/netns/hash.h> 35 #include <net/lwtunnel.h> 36 37 #define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */ 38 #define IPV4_MIN_MTU 68 /* RFC 791 */ 39 40 extern unsigned int sysctl_fib_sync_mem; 41 extern unsigned int sysctl_fib_sync_mem_min; 42 extern unsigned int sysctl_fib_sync_mem_max; 43 44 struct sock; 45 46 struct inet_skb_parm { 47 int iif; 48 struct ip_options opt; /* Compiled IP options */ 49 u16 flags; 50 51 #define IPSKB_FORWARDED BIT(0) 52 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1) 53 #define IPSKB_XFRM_TRANSFORMED BIT(2) 54 #define IPSKB_FRAG_COMPLETE BIT(3) 55 #define IPSKB_REROUTED BIT(4) 56 #define IPSKB_DOREDIRECT BIT(5) 57 #define IPSKB_FRAG_PMTU BIT(6) 58 #define IPSKB_L3SLAVE BIT(7) 59 #define IPSKB_NOPOLICY BIT(8) 60 61 u16 frag_max_size; 62 }; 63 64 static inline bool ipv4_l3mdev_skb(u16 flags) 65 { 66 return !!(flags & IPSKB_L3SLAVE); 67 } 68 69 static inline unsigned int ip_hdrlen(const struct sk_buff *skb) 70 { 71 return ip_hdr(skb)->ihl * 4; 72 } 73 74 struct ipcm_cookie { 75 struct sockcm_cookie sockc; 76 __be32 addr; 77 int oif; 78 struct ip_options_rcu *opt; 79 __u8 ttl; 80 __s16 tos; 81 char priority; 82 __u16 gso_size; 83 }; 84 85 static inline void ipcm_init(struct ipcm_cookie *ipcm) 86 { 87 *ipcm = (struct ipcm_cookie) { .tos = -1 }; 88 } 89 90 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm, 91 const struct inet_sock *inet) 92 { 93 ipcm_init(ipcm); 94 95 ipcm->sockc.mark = inet->sk.sk_mark; 96 ipcm->sockc.tsflags = inet->sk.sk_tsflags; 97 ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if); 98 ipcm->addr = inet->inet_saddr; 99 } 100 101 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) 102 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) 103 104 /* return enslaved device index if relevant */ 105 static inline int inet_sdif(const struct sk_buff *skb) 106 { 107 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV) 108 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags)) 109 return IPCB(skb)->iif; 110 #endif 111 return 0; 112 } 113 114 /* Special input handler for packets caught by router alert option. 115 They are selected only by protocol field, and then processed likely 116 local ones; but only if someone wants them! Otherwise, router 117 not running rsvpd will kill RSVP. 118 119 It is user level problem, what it will make with them. 120 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), 121 but receiver should be enough clever f.e. to forward mtrace requests, 122 sent to multicast group to reach destination designated router. 123 */ 124 125 struct ip_ra_chain { 126 struct ip_ra_chain __rcu *next; 127 struct sock *sk; 128 union { 129 void (*destructor)(struct sock *); 130 struct sock *saved_sk; 131 }; 132 struct rcu_head rcu; 133 }; 134 135 /* IP flags. */ 136 #define IP_CE 0x8000 /* Flag: "Congestion" */ 137 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */ 138 #define IP_MF 0x2000 /* Flag: "More Fragments" */ 139 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ 140 141 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ 142 143 struct msghdr; 144 struct net_device; 145 struct packet_type; 146 struct rtable; 147 struct sockaddr; 148 149 int igmp_mc_init(void); 150 151 /* 152 * Functions provided by ip.c 153 */ 154 155 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk, 156 __be32 saddr, __be32 daddr, 157 struct ip_options_rcu *opt, u8 tos); 158 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, 159 struct net_device *orig_dev); 160 void ip_list_rcv(struct list_head *head, struct packet_type *pt, 161 struct net_device *orig_dev); 162 int ip_local_deliver(struct sk_buff *skb); 163 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto); 164 int ip_mr_input(struct sk_buff *skb); 165 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb); 166 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb); 167 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, 168 int (*output)(struct net *, struct sock *, struct sk_buff *)); 169 170 struct ip_fraglist_iter { 171 struct sk_buff *frag; 172 struct iphdr *iph; 173 int offset; 174 unsigned int hlen; 175 }; 176 177 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph, 178 unsigned int hlen, struct ip_fraglist_iter *iter); 179 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter); 180 181 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter) 182 { 183 struct sk_buff *skb = iter->frag; 184 185 iter->frag = skb->next; 186 skb_mark_not_on_list(skb); 187 188 return skb; 189 } 190 191 struct ip_frag_state { 192 bool DF; 193 unsigned int hlen; 194 unsigned int ll_rs; 195 unsigned int mtu; 196 unsigned int left; 197 int offset; 198 int ptr; 199 __be16 not_last_frag; 200 }; 201 202 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs, 203 unsigned int mtu, bool DF, struct ip_frag_state *state); 204 struct sk_buff *ip_frag_next(struct sk_buff *skb, 205 struct ip_frag_state *state); 206 207 void ip_send_check(struct iphdr *ip); 208 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 209 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 210 211 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl, 212 __u8 tos); 213 void ip_init(void); 214 int ip_append_data(struct sock *sk, struct flowi4 *fl4, 215 int getfrag(void *from, char *to, int offset, int len, 216 int odd, struct sk_buff *skb), 217 void *from, int len, int protolen, 218 struct ipcm_cookie *ipc, 219 struct rtable **rt, 220 unsigned int flags); 221 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, 222 struct sk_buff *skb); 223 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, 224 struct sk_buff_head *queue, 225 struct inet_cork *cork); 226 int ip_send_skb(struct net *net, struct sk_buff *skb); 227 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); 228 void ip_flush_pending_frames(struct sock *sk); 229 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, 230 int getfrag(void *from, char *to, int offset, 231 int len, int odd, struct sk_buff *skb), 232 void *from, int length, int transhdrlen, 233 struct ipcm_cookie *ipc, struct rtable **rtp, 234 struct inet_cork *cork, unsigned int flags); 235 236 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); 237 238 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) 239 { 240 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base); 241 } 242 243 /* Get the route scope that should be used when sending a packet. */ 244 static inline u8 ip_sendmsg_scope(const struct inet_sock *inet, 245 const struct ipcm_cookie *ipc, 246 const struct msghdr *msg) 247 { 248 if (sock_flag(&inet->sk, SOCK_LOCALROUTE) || 249 msg->msg_flags & MSG_DONTROUTE || 250 (ipc->opt && ipc->opt->opt.is_strictroute)) 251 return RT_SCOPE_LINK; 252 253 return RT_SCOPE_UNIVERSE; 254 } 255 256 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet) 257 { 258 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos); 259 } 260 261 /* datagram.c */ 262 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 263 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 264 265 void ip4_datagram_release_cb(struct sock *sk); 266 267 struct ip_reply_arg { 268 struct kvec iov[1]; 269 int flags; 270 __wsum csum; 271 int csumoffset; /* u16 offset of csum in iov[0].iov_base */ 272 /* -1 if not needed */ 273 int bound_dev_if; 274 u8 tos; 275 kuid_t uid; 276 }; 277 278 #define IP_REPLY_ARG_NOSRCCHECK 1 279 280 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg) 281 { 282 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0; 283 } 284 285 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, 286 const struct ip_options *sopt, 287 __be32 daddr, __be32 saddr, 288 const struct ip_reply_arg *arg, 289 unsigned int len, u64 transmit_time, u32 txhash); 290 291 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) 292 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field) 293 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 294 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 295 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 296 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 297 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) 298 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field) 299 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 300 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 301 302 static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt) 303 { 304 return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt); 305 } 306 307 unsigned long snmp_fold_field(void __percpu *mib, int offt); 308 #if BITS_PER_LONG==32 309 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 310 size_t syncp_offset); 311 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); 312 #else 313 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 314 size_t syncp_offset) 315 { 316 return snmp_get_cpu_field(mib, cpu, offct); 317 318 } 319 320 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) 321 { 322 return snmp_fold_field(mib, offt); 323 } 324 #endif 325 326 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \ 327 { \ 328 int i, c; \ 329 for_each_possible_cpu(c) { \ 330 for (i = 0; stats_list[i].name; i++) \ 331 buff64[i] += snmp_get_cpu_field64( \ 332 mib_statistic, \ 333 c, stats_list[i].entry, \ 334 offset); \ 335 } \ 336 } 337 338 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \ 339 { \ 340 int i, c; \ 341 for_each_possible_cpu(c) { \ 342 for (i = 0; stats_list[i].name; i++) \ 343 buff[i] += snmp_get_cpu_field( \ 344 mib_statistic, \ 345 c, stats_list[i].entry); \ 346 } \ 347 } 348 349 void inet_get_local_port_range(const struct net *net, int *low, int *high); 350 void inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high); 351 352 #ifdef CONFIG_SYSCTL 353 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) 354 { 355 if (!net->ipv4.sysctl_local_reserved_ports) 356 return false; 357 return test_bit(port, net->ipv4.sysctl_local_reserved_ports); 358 } 359 360 static inline bool sysctl_dev_name_is_allowed(const char *name) 361 { 362 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; 363 } 364 365 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) 366 { 367 return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock); 368 } 369 370 #else 371 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) 372 { 373 return false; 374 } 375 376 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) 377 { 378 return port < PROT_SOCK; 379 } 380 #endif 381 382 __be32 inet_current_timestamp(void); 383 384 /* From inetpeer.c */ 385 extern int inet_peer_threshold; 386 extern int inet_peer_minttl; 387 extern int inet_peer_maxttl; 388 389 void ipfrag_init(void); 390 391 void ip_static_sysctl_init(void); 392 393 #define IP4_REPLY_MARK(net, mark) \ 394 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0) 395 396 static inline bool ip_is_fragment(const struct iphdr *iph) 397 { 398 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; 399 } 400 401 #ifdef CONFIG_INET 402 #include <net/dst.h> 403 404 /* The function in 2.2 was invalid, producing wrong result for 405 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ 406 static inline 407 int ip_decrease_ttl(struct iphdr *iph) 408 { 409 u32 check = (__force u32)iph->check; 410 check += (__force u32)htons(0x0100); 411 iph->check = (__force __sum16)(check + (check>=0xFFFF)); 412 return --iph->ttl; 413 } 414 415 static inline int ip_mtu_locked(const struct dst_entry *dst) 416 { 417 const struct rtable *rt = (const struct rtable *)dst; 418 419 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU); 420 } 421 422 static inline 423 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst) 424 { 425 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); 426 427 return pmtudisc == IP_PMTUDISC_DO || 428 (pmtudisc == IP_PMTUDISC_WANT && 429 !ip_mtu_locked(dst)); 430 } 431 432 static inline bool ip_sk_accept_pmtu(const struct sock *sk) 433 { 434 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE && 435 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT; 436 } 437 438 static inline bool ip_sk_use_pmtu(const struct sock *sk) 439 { 440 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE; 441 } 442 443 static inline bool ip_sk_ignore_df(const struct sock *sk) 444 { 445 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO || 446 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT; 447 } 448 449 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, 450 bool forwarding) 451 { 452 const struct rtable *rt = container_of(dst, struct rtable, dst); 453 struct net *net = dev_net(dst->dev); 454 unsigned int mtu; 455 456 if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) || 457 ip_mtu_locked(dst) || 458 !forwarding) { 459 mtu = rt->rt_pmtu; 460 if (mtu && time_before(jiffies, rt->dst.expires)) 461 goto out; 462 } 463 464 /* 'forwarding = true' case should always honour route mtu */ 465 mtu = dst_metric_raw(dst, RTAX_MTU); 466 if (mtu) 467 goto out; 468 469 mtu = READ_ONCE(dst->dev->mtu); 470 471 if (unlikely(ip_mtu_locked(dst))) { 472 if (rt->rt_uses_gateway && mtu > 576) 473 mtu = 576; 474 } 475 476 out: 477 mtu = min_t(unsigned int, mtu, IP_MAX_MTU); 478 479 return mtu - lwtunnel_headroom(dst->lwtstate, mtu); 480 } 481 482 static inline unsigned int ip_skb_dst_mtu(struct sock *sk, 483 const struct sk_buff *skb) 484 { 485 unsigned int mtu; 486 487 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) { 488 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; 489 490 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding); 491 } 492 493 mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU); 494 return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu); 495 } 496 497 struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx, 498 int fc_mx_len, 499 struct netlink_ext_ack *extack); 500 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics) 501 { 502 if (fib_metrics != &dst_default_metrics && 503 refcount_dec_and_test(&fib_metrics->refcnt)) 504 kfree(fib_metrics); 505 } 506 507 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */ 508 static inline 509 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics) 510 { 511 dst_init_metrics(dst, fib_metrics->metrics, true); 512 513 if (fib_metrics != &dst_default_metrics) { 514 dst->_metrics |= DST_METRICS_REFCOUNTED; 515 refcount_inc(&fib_metrics->refcnt); 516 } 517 } 518 519 static inline 520 void ip_dst_metrics_put(struct dst_entry *dst) 521 { 522 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst); 523 524 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt)) 525 kfree(p); 526 } 527 528 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); 529 530 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, 531 struct sock *sk, int segs) 532 { 533 struct iphdr *iph = ip_hdr(skb); 534 535 /* We had many attacks based on IPID, use the private 536 * generator as much as we can. 537 */ 538 if (sk && inet_sk(sk)->inet_daddr) { 539 iph->id = htons(inet_sk(sk)->inet_id); 540 inet_sk(sk)->inet_id += segs; 541 return; 542 } 543 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { 544 iph->id = 0; 545 } else { 546 /* Unfortunately we need the big hammer to get a suitable IPID */ 547 __ip_select_ident(net, iph, segs); 548 } 549 } 550 551 static inline void ip_select_ident(struct net *net, struct sk_buff *skb, 552 struct sock *sk) 553 { 554 ip_select_ident_segs(net, skb, sk, 1); 555 } 556 557 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) 558 { 559 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, 560 skb->len, proto, 0); 561 } 562 563 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store 564 * Equivalent to : flow->v4addrs.src = iph->saddr; 565 * flow->v4addrs.dst = iph->daddr; 566 */ 567 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, 568 const struct iphdr *iph) 569 { 570 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != 571 offsetof(typeof(flow->addrs), v4addrs.src) + 572 sizeof(flow->addrs.v4addrs.src)); 573 memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs)); 574 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; 575 } 576 577 /* 578 * Map a multicast IP onto multicast MAC for type ethernet. 579 */ 580 581 static inline void ip_eth_mc_map(__be32 naddr, char *buf) 582 { 583 __u32 addr=ntohl(naddr); 584 buf[0]=0x01; 585 buf[1]=0x00; 586 buf[2]=0x5e; 587 buf[5]=addr&0xFF; 588 addr>>=8; 589 buf[4]=addr&0xFF; 590 addr>>=8; 591 buf[3]=addr&0x7F; 592 } 593 594 /* 595 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. 596 * Leave P_Key as 0 to be filled in by driver. 597 */ 598 599 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 600 { 601 __u32 addr; 602 unsigned char scope = broadcast[5] & 0xF; 603 604 buf[0] = 0; /* Reserved */ 605 buf[1] = 0xff; /* Multicast QPN */ 606 buf[2] = 0xff; 607 buf[3] = 0xff; 608 addr = ntohl(naddr); 609 buf[4] = 0xff; 610 buf[5] = 0x10 | scope; /* scope from broadcast address */ 611 buf[6] = 0x40; /* IPv4 signature */ 612 buf[7] = 0x1b; 613 buf[8] = broadcast[8]; /* P_Key */ 614 buf[9] = broadcast[9]; 615 buf[10] = 0; 616 buf[11] = 0; 617 buf[12] = 0; 618 buf[13] = 0; 619 buf[14] = 0; 620 buf[15] = 0; 621 buf[19] = addr & 0xff; 622 addr >>= 8; 623 buf[18] = addr & 0xff; 624 addr >>= 8; 625 buf[17] = addr & 0xff; 626 addr >>= 8; 627 buf[16] = addr & 0x0f; 628 } 629 630 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 631 { 632 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) 633 memcpy(buf, broadcast, 4); 634 else 635 memcpy(buf, &naddr, sizeof(naddr)); 636 } 637 638 #if IS_ENABLED(CONFIG_IPV6) 639 #include <linux/ipv6.h> 640 #endif 641 642 static __inline__ void inet_reset_saddr(struct sock *sk) 643 { 644 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; 645 #if IS_ENABLED(CONFIG_IPV6) 646 if (sk->sk_family == PF_INET6) { 647 struct ipv6_pinfo *np = inet6_sk(sk); 648 649 memset(&np->saddr, 0, sizeof(np->saddr)); 650 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); 651 } 652 #endif 653 } 654 655 #endif 656 657 static inline unsigned int ipv4_addr_hash(__be32 ip) 658 { 659 return (__force unsigned int) ip; 660 } 661 662 static inline u32 ipv4_portaddr_hash(const struct net *net, 663 __be32 saddr, 664 unsigned int port) 665 { 666 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port; 667 } 668 669 bool ip_call_ra_chain(struct sk_buff *skb); 670 671 /* 672 * Functions provided by ip_fragment.c 673 */ 674 675 enum ip_defrag_users { 676 IP_DEFRAG_LOCAL_DELIVER, 677 IP_DEFRAG_CALL_RA_CHAIN, 678 IP_DEFRAG_CONNTRACK_IN, 679 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, 680 IP_DEFRAG_CONNTRACK_OUT, 681 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 682 IP_DEFRAG_CONNTRACK_BRIDGE_IN, 683 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 684 IP_DEFRAG_VS_IN, 685 IP_DEFRAG_VS_OUT, 686 IP_DEFRAG_VS_FWD, 687 IP_DEFRAG_AF_PACKET, 688 IP_DEFRAG_MACVLAN, 689 }; 690 691 /* Return true if the value of 'user' is between 'lower_bond' 692 * and 'upper_bond' inclusively. 693 */ 694 static inline bool ip_defrag_user_in_between(u32 user, 695 enum ip_defrag_users lower_bond, 696 enum ip_defrag_users upper_bond) 697 { 698 return user >= lower_bond && user <= upper_bond; 699 } 700 701 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user); 702 #ifdef CONFIG_INET 703 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user); 704 #else 705 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user) 706 { 707 return skb; 708 } 709 #endif 710 711 /* 712 * Functions provided by ip_forward.c 713 */ 714 715 int ip_forward(struct sk_buff *skb); 716 717 /* 718 * Functions provided by ip_options.c 719 */ 720 721 void ip_options_build(struct sk_buff *skb, struct ip_options *opt, 722 __be32 daddr, struct rtable *rt); 723 724 int __ip_options_echo(struct net *net, struct ip_options *dopt, 725 struct sk_buff *skb, const struct ip_options *sopt); 726 static inline int ip_options_echo(struct net *net, struct ip_options *dopt, 727 struct sk_buff *skb) 728 { 729 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt); 730 } 731 732 void ip_options_fragment(struct sk_buff *skb); 733 int __ip_options_compile(struct net *net, struct ip_options *opt, 734 struct sk_buff *skb, __be32 *info); 735 int ip_options_compile(struct net *net, struct ip_options *opt, 736 struct sk_buff *skb); 737 int ip_options_get(struct net *net, struct ip_options_rcu **optp, 738 sockptr_t data, int optlen); 739 void ip_options_undo(struct ip_options *opt); 740 void ip_forward_options(struct sk_buff *skb); 741 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev); 742 743 /* 744 * Functions provided by ip_sockglue.c 745 */ 746 747 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb); 748 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk, 749 struct sk_buff *skb, int tlen, int offset); 750 int ip_cmsg_send(struct sock *sk, struct msghdr *msg, 751 struct ipcm_cookie *ipc, bool allow_ipv6); 752 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl); 753 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, 754 unsigned int optlen); 755 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, 756 unsigned int optlen); 757 int do_ip_getsockopt(struct sock *sk, int level, int optname, 758 sockptr_t optval, sockptr_t optlen); 759 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, 760 int __user *optlen); 761 int ip_ra_control(struct sock *sk, unsigned char on, 762 void (*destructor)(struct sock *)); 763 764 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); 765 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 766 u32 info, u8 *payload); 767 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, 768 u32 info); 769 770 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 771 { 772 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0); 773 } 774 775 bool icmp_global_allow(void); 776 extern int sysctl_icmp_msgs_per_sec; 777 extern int sysctl_icmp_msgs_burst; 778 779 #ifdef CONFIG_PROC_FS 780 int ip_misc_proc_init(void); 781 #endif 782 783 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family, 784 struct netlink_ext_ack *extack); 785 786 static inline bool inetdev_valid_mtu(unsigned int mtu) 787 { 788 return likely(mtu >= IPV4_MIN_MTU); 789 } 790 791 void ip_sock_set_freebind(struct sock *sk); 792 int ip_sock_set_mtu_discover(struct sock *sk, int val); 793 void ip_sock_set_pktinfo(struct sock *sk); 794 void ip_sock_set_recverr(struct sock *sk); 795 void ip_sock_set_tos(struct sock *sk, int val); 796 void __ip_sock_set_tos(struct sock *sk, int val); 797 798 #endif /* _IP_H */ 799