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 protocol; 80 __u8 ttl; 81 __s16 tos; 82 char priority; 83 __u16 gso_size; 84 }; 85 86 static inline void ipcm_init(struct ipcm_cookie *ipcm) 87 { 88 *ipcm = (struct ipcm_cookie) { .tos = -1 }; 89 } 90 91 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm, 92 const struct inet_sock *inet) 93 { 94 ipcm_init(ipcm); 95 96 ipcm->sockc.mark = READ_ONCE(inet->sk.sk_mark); 97 ipcm->sockc.tsflags = inet->sk.sk_tsflags; 98 ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if); 99 ipcm->addr = inet->inet_saddr; 100 ipcm->protocol = inet->inet_num; 101 } 102 103 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) 104 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) 105 106 /* return enslaved device index if relevant */ 107 static inline int inet_sdif(const struct sk_buff *skb) 108 { 109 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV) 110 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags)) 111 return IPCB(skb)->iif; 112 #endif 113 return 0; 114 } 115 116 /* Special input handler for packets caught by router alert option. 117 They are selected only by protocol field, and then processed likely 118 local ones; but only if someone wants them! Otherwise, router 119 not running rsvpd will kill RSVP. 120 121 It is user level problem, what it will make with them. 122 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), 123 but receiver should be enough clever f.e. to forward mtrace requests, 124 sent to multicast group to reach destination designated router. 125 */ 126 127 struct ip_ra_chain { 128 struct ip_ra_chain __rcu *next; 129 struct sock *sk; 130 union { 131 void (*destructor)(struct sock *); 132 struct sock *saved_sk; 133 }; 134 struct rcu_head rcu; 135 }; 136 137 /* IP flags. */ 138 #define IP_CE 0x8000 /* Flag: "Congestion" */ 139 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */ 140 #define IP_MF 0x2000 /* Flag: "More Fragments" */ 141 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ 142 143 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ 144 145 struct msghdr; 146 struct net_device; 147 struct packet_type; 148 struct rtable; 149 struct sockaddr; 150 151 int igmp_mc_init(void); 152 153 /* 154 * Functions provided by ip.c 155 */ 156 157 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk, 158 __be32 saddr, __be32 daddr, 159 struct ip_options_rcu *opt, u8 tos); 160 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, 161 struct net_device *orig_dev); 162 void ip_list_rcv(struct list_head *head, struct packet_type *pt, 163 struct net_device *orig_dev); 164 int ip_local_deliver(struct sk_buff *skb); 165 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto); 166 int ip_mr_input(struct sk_buff *skb); 167 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb); 168 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb); 169 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, 170 int (*output)(struct net *, struct sock *, struct sk_buff *)); 171 172 struct ip_fraglist_iter { 173 struct sk_buff *frag; 174 struct iphdr *iph; 175 int offset; 176 unsigned int hlen; 177 }; 178 179 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph, 180 unsigned int hlen, struct ip_fraglist_iter *iter); 181 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter); 182 183 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter) 184 { 185 struct sk_buff *skb = iter->frag; 186 187 iter->frag = skb->next; 188 skb_mark_not_on_list(skb); 189 190 return skb; 191 } 192 193 struct ip_frag_state { 194 bool DF; 195 unsigned int hlen; 196 unsigned int ll_rs; 197 unsigned int mtu; 198 unsigned int left; 199 int offset; 200 int ptr; 201 __be16 not_last_frag; 202 }; 203 204 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs, 205 unsigned int mtu, bool DF, struct ip_frag_state *state); 206 struct sk_buff *ip_frag_next(struct sk_buff *skb, 207 struct ip_frag_state *state); 208 209 void ip_send_check(struct iphdr *ip); 210 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 211 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 212 213 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl, 214 __u8 tos); 215 void ip_init(void); 216 int ip_append_data(struct sock *sk, struct flowi4 *fl4, 217 int getfrag(void *from, char *to, int offset, int len, 218 int odd, struct sk_buff *skb), 219 void *from, int len, int protolen, 220 struct ipcm_cookie *ipc, 221 struct rtable **rt, 222 unsigned int flags); 223 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, 224 struct sk_buff *skb); 225 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, 226 struct sk_buff_head *queue, 227 struct inet_cork *cork); 228 int ip_send_skb(struct net *net, struct sk_buff *skb); 229 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); 230 void ip_flush_pending_frames(struct sock *sk); 231 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, 232 int getfrag(void *from, char *to, int offset, 233 int len, int odd, struct sk_buff *skb), 234 void *from, int length, int transhdrlen, 235 struct ipcm_cookie *ipc, struct rtable **rtp, 236 struct inet_cork *cork, unsigned int flags); 237 238 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); 239 240 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) 241 { 242 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base); 243 } 244 245 /* Get the route scope that should be used when sending a packet. */ 246 static inline u8 ip_sendmsg_scope(const struct inet_sock *inet, 247 const struct ipcm_cookie *ipc, 248 const struct msghdr *msg) 249 { 250 if (sock_flag(&inet->sk, SOCK_LOCALROUTE) || 251 msg->msg_flags & MSG_DONTROUTE || 252 (ipc->opt && ipc->opt->opt.is_strictroute)) 253 return RT_SCOPE_LINK; 254 255 return RT_SCOPE_UNIVERSE; 256 } 257 258 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet) 259 { 260 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos); 261 } 262 263 /* datagram.c */ 264 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 265 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); 266 267 void ip4_datagram_release_cb(struct sock *sk); 268 269 struct ip_reply_arg { 270 struct kvec iov[1]; 271 int flags; 272 __wsum csum; 273 int csumoffset; /* u16 offset of csum in iov[0].iov_base */ 274 /* -1 if not needed */ 275 int bound_dev_if; 276 u8 tos; 277 kuid_t uid; 278 }; 279 280 #define IP_REPLY_ARG_NOSRCCHECK 1 281 282 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg) 283 { 284 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0; 285 } 286 287 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb, 288 const struct ip_options *sopt, 289 __be32 daddr, __be32 saddr, 290 const struct ip_reply_arg *arg, 291 unsigned int len, u64 transmit_time, u32 txhash); 292 293 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) 294 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field) 295 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 296 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 297 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 298 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 299 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) 300 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field) 301 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 302 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 303 304 static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt) 305 { 306 return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt); 307 } 308 309 unsigned long snmp_fold_field(void __percpu *mib, int offt); 310 #if BITS_PER_LONG==32 311 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 312 size_t syncp_offset); 313 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); 314 #else 315 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 316 size_t syncp_offset) 317 { 318 return snmp_get_cpu_field(mib, cpu, offct); 319 320 } 321 322 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) 323 { 324 return snmp_fold_field(mib, offt); 325 } 326 #endif 327 328 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \ 329 { \ 330 int i, c; \ 331 for_each_possible_cpu(c) { \ 332 for (i = 0; stats_list[i].name; i++) \ 333 buff64[i] += snmp_get_cpu_field64( \ 334 mib_statistic, \ 335 c, stats_list[i].entry, \ 336 offset); \ 337 } \ 338 } 339 340 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \ 341 { \ 342 int i, c; \ 343 for_each_possible_cpu(c) { \ 344 for (i = 0; stats_list[i].name; i++) \ 345 buff[i] += snmp_get_cpu_field( \ 346 mib_statistic, \ 347 c, stats_list[i].entry); \ 348 } \ 349 } 350 351 void inet_get_local_port_range(const struct net *net, int *low, int *high); 352 void inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high); 353 354 #ifdef CONFIG_SYSCTL 355 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) 356 { 357 if (!net->ipv4.sysctl_local_reserved_ports) 358 return false; 359 return test_bit(port, net->ipv4.sysctl_local_reserved_ports); 360 } 361 362 static inline bool sysctl_dev_name_is_allowed(const char *name) 363 { 364 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; 365 } 366 367 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) 368 { 369 return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock); 370 } 371 372 #else 373 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) 374 { 375 return false; 376 } 377 378 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) 379 { 380 return port < PROT_SOCK; 381 } 382 #endif 383 384 __be32 inet_current_timestamp(void); 385 386 /* From inetpeer.c */ 387 extern int inet_peer_threshold; 388 extern int inet_peer_minttl; 389 extern int inet_peer_maxttl; 390 391 void ipfrag_init(void); 392 393 void ip_static_sysctl_init(void); 394 395 #define IP4_REPLY_MARK(net, mark) \ 396 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0) 397 398 static inline bool ip_is_fragment(const struct iphdr *iph) 399 { 400 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; 401 } 402 403 #ifdef CONFIG_INET 404 #include <net/dst.h> 405 406 /* The function in 2.2 was invalid, producing wrong result for 407 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ 408 static inline 409 int ip_decrease_ttl(struct iphdr *iph) 410 { 411 u32 check = (__force u32)iph->check; 412 check += (__force u32)htons(0x0100); 413 iph->check = (__force __sum16)(check + (check>=0xFFFF)); 414 return --iph->ttl; 415 } 416 417 static inline int ip_mtu_locked(const struct dst_entry *dst) 418 { 419 const struct rtable *rt = (const struct rtable *)dst; 420 421 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU); 422 } 423 424 static inline 425 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst) 426 { 427 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); 428 429 return pmtudisc == IP_PMTUDISC_DO || 430 (pmtudisc == IP_PMTUDISC_WANT && 431 !ip_mtu_locked(dst)); 432 } 433 434 static inline bool ip_sk_accept_pmtu(const struct sock *sk) 435 { 436 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE && 437 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT; 438 } 439 440 static inline bool ip_sk_use_pmtu(const struct sock *sk) 441 { 442 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE; 443 } 444 445 static inline bool ip_sk_ignore_df(const struct sock *sk) 446 { 447 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO || 448 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT; 449 } 450 451 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, 452 bool forwarding) 453 { 454 const struct rtable *rt = container_of(dst, struct rtable, dst); 455 struct net *net = dev_net(dst->dev); 456 unsigned int mtu; 457 458 if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) || 459 ip_mtu_locked(dst) || 460 !forwarding) { 461 mtu = rt->rt_pmtu; 462 if (mtu && time_before(jiffies, rt->dst.expires)) 463 goto out; 464 } 465 466 /* 'forwarding = true' case should always honour route mtu */ 467 mtu = dst_metric_raw(dst, RTAX_MTU); 468 if (mtu) 469 goto out; 470 471 mtu = READ_ONCE(dst->dev->mtu); 472 473 if (unlikely(ip_mtu_locked(dst))) { 474 if (rt->rt_uses_gateway && mtu > 576) 475 mtu = 576; 476 } 477 478 out: 479 mtu = min_t(unsigned int, mtu, IP_MAX_MTU); 480 481 return mtu - lwtunnel_headroom(dst->lwtstate, mtu); 482 } 483 484 static inline unsigned int ip_skb_dst_mtu(struct sock *sk, 485 const struct sk_buff *skb) 486 { 487 unsigned int mtu; 488 489 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) { 490 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; 491 492 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding); 493 } 494 495 mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU); 496 return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu); 497 } 498 499 struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx, 500 int fc_mx_len, 501 struct netlink_ext_ack *extack); 502 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics) 503 { 504 if (fib_metrics != &dst_default_metrics && 505 refcount_dec_and_test(&fib_metrics->refcnt)) 506 kfree(fib_metrics); 507 } 508 509 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */ 510 static inline 511 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics) 512 { 513 dst_init_metrics(dst, fib_metrics->metrics, true); 514 515 if (fib_metrics != &dst_default_metrics) { 516 dst->_metrics |= DST_METRICS_REFCOUNTED; 517 refcount_inc(&fib_metrics->refcnt); 518 } 519 } 520 521 static inline 522 void ip_dst_metrics_put(struct dst_entry *dst) 523 { 524 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst); 525 526 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt)) 527 kfree(p); 528 } 529 530 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); 531 532 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, 533 struct sock *sk, int segs) 534 { 535 struct iphdr *iph = ip_hdr(skb); 536 537 /* We had many attacks based on IPID, use the private 538 * generator as much as we can. 539 */ 540 if (sk && inet_sk(sk)->inet_daddr) { 541 int val; 542 543 /* avoid atomic operations for TCP, 544 * as we hold socket lock at this point. 545 */ 546 if (sk_is_tcp(sk)) { 547 sock_owned_by_me(sk); 548 val = atomic_read(&inet_sk(sk)->inet_id); 549 atomic_set(&inet_sk(sk)->inet_id, val + segs); 550 } else { 551 val = atomic_add_return(segs, &inet_sk(sk)->inet_id); 552 } 553 iph->id = htons(val); 554 return; 555 } 556 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { 557 iph->id = 0; 558 } else { 559 /* Unfortunately we need the big hammer to get a suitable IPID */ 560 __ip_select_ident(net, iph, segs); 561 } 562 } 563 564 static inline void ip_select_ident(struct net *net, struct sk_buff *skb, 565 struct sock *sk) 566 { 567 ip_select_ident_segs(net, skb, sk, 1); 568 } 569 570 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) 571 { 572 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, 573 skb->len, proto, 0); 574 } 575 576 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store 577 * Equivalent to : flow->v4addrs.src = iph->saddr; 578 * flow->v4addrs.dst = iph->daddr; 579 */ 580 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, 581 const struct iphdr *iph) 582 { 583 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != 584 offsetof(typeof(flow->addrs), v4addrs.src) + 585 sizeof(flow->addrs.v4addrs.src)); 586 memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs)); 587 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; 588 } 589 590 /* 591 * Map a multicast IP onto multicast MAC for type ethernet. 592 */ 593 594 static inline void ip_eth_mc_map(__be32 naddr, char *buf) 595 { 596 __u32 addr=ntohl(naddr); 597 buf[0]=0x01; 598 buf[1]=0x00; 599 buf[2]=0x5e; 600 buf[5]=addr&0xFF; 601 addr>>=8; 602 buf[4]=addr&0xFF; 603 addr>>=8; 604 buf[3]=addr&0x7F; 605 } 606 607 /* 608 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. 609 * Leave P_Key as 0 to be filled in by driver. 610 */ 611 612 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 613 { 614 __u32 addr; 615 unsigned char scope = broadcast[5] & 0xF; 616 617 buf[0] = 0; /* Reserved */ 618 buf[1] = 0xff; /* Multicast QPN */ 619 buf[2] = 0xff; 620 buf[3] = 0xff; 621 addr = ntohl(naddr); 622 buf[4] = 0xff; 623 buf[5] = 0x10 | scope; /* scope from broadcast address */ 624 buf[6] = 0x40; /* IPv4 signature */ 625 buf[7] = 0x1b; 626 buf[8] = broadcast[8]; /* P_Key */ 627 buf[9] = broadcast[9]; 628 buf[10] = 0; 629 buf[11] = 0; 630 buf[12] = 0; 631 buf[13] = 0; 632 buf[14] = 0; 633 buf[15] = 0; 634 buf[19] = addr & 0xff; 635 addr >>= 8; 636 buf[18] = addr & 0xff; 637 addr >>= 8; 638 buf[17] = addr & 0xff; 639 addr >>= 8; 640 buf[16] = addr & 0x0f; 641 } 642 643 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 644 { 645 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) 646 memcpy(buf, broadcast, 4); 647 else 648 memcpy(buf, &naddr, sizeof(naddr)); 649 } 650 651 #if IS_ENABLED(CONFIG_IPV6) 652 #include <linux/ipv6.h> 653 #endif 654 655 static __inline__ void inet_reset_saddr(struct sock *sk) 656 { 657 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; 658 #if IS_ENABLED(CONFIG_IPV6) 659 if (sk->sk_family == PF_INET6) { 660 struct ipv6_pinfo *np = inet6_sk(sk); 661 662 memset(&np->saddr, 0, sizeof(np->saddr)); 663 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); 664 } 665 #endif 666 } 667 668 #endif 669 670 static inline unsigned int ipv4_addr_hash(__be32 ip) 671 { 672 return (__force unsigned int) ip; 673 } 674 675 static inline u32 ipv4_portaddr_hash(const struct net *net, 676 __be32 saddr, 677 unsigned int port) 678 { 679 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port; 680 } 681 682 bool ip_call_ra_chain(struct sk_buff *skb); 683 684 /* 685 * Functions provided by ip_fragment.c 686 */ 687 688 enum ip_defrag_users { 689 IP_DEFRAG_LOCAL_DELIVER, 690 IP_DEFRAG_CALL_RA_CHAIN, 691 IP_DEFRAG_CONNTRACK_IN, 692 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, 693 IP_DEFRAG_CONNTRACK_OUT, 694 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 695 IP_DEFRAG_CONNTRACK_BRIDGE_IN, 696 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 697 IP_DEFRAG_VS_IN, 698 IP_DEFRAG_VS_OUT, 699 IP_DEFRAG_VS_FWD, 700 IP_DEFRAG_AF_PACKET, 701 IP_DEFRAG_MACVLAN, 702 }; 703 704 /* Return true if the value of 'user' is between 'lower_bond' 705 * and 'upper_bond' inclusively. 706 */ 707 static inline bool ip_defrag_user_in_between(u32 user, 708 enum ip_defrag_users lower_bond, 709 enum ip_defrag_users upper_bond) 710 { 711 return user >= lower_bond && user <= upper_bond; 712 } 713 714 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user); 715 #ifdef CONFIG_INET 716 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user); 717 #else 718 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user) 719 { 720 return skb; 721 } 722 #endif 723 724 /* 725 * Functions provided by ip_forward.c 726 */ 727 728 int ip_forward(struct sk_buff *skb); 729 730 /* 731 * Functions provided by ip_options.c 732 */ 733 734 void ip_options_build(struct sk_buff *skb, struct ip_options *opt, 735 __be32 daddr, struct rtable *rt); 736 737 int __ip_options_echo(struct net *net, struct ip_options *dopt, 738 struct sk_buff *skb, const struct ip_options *sopt); 739 static inline int ip_options_echo(struct net *net, struct ip_options *dopt, 740 struct sk_buff *skb) 741 { 742 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt); 743 } 744 745 void ip_options_fragment(struct sk_buff *skb); 746 int __ip_options_compile(struct net *net, struct ip_options *opt, 747 struct sk_buff *skb, __be32 *info); 748 int ip_options_compile(struct net *net, struct ip_options *opt, 749 struct sk_buff *skb); 750 int ip_options_get(struct net *net, struct ip_options_rcu **optp, 751 sockptr_t data, int optlen); 752 void ip_options_undo(struct ip_options *opt); 753 void ip_forward_options(struct sk_buff *skb); 754 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev); 755 756 /* 757 * Functions provided by ip_sockglue.c 758 */ 759 760 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb); 761 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk, 762 struct sk_buff *skb, int tlen, int offset); 763 int ip_cmsg_send(struct sock *sk, struct msghdr *msg, 764 struct ipcm_cookie *ipc, bool allow_ipv6); 765 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl); 766 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, 767 unsigned int optlen); 768 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, 769 unsigned int optlen); 770 int do_ip_getsockopt(struct sock *sk, int level, int optname, 771 sockptr_t optval, sockptr_t optlen); 772 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, 773 int __user *optlen); 774 int ip_ra_control(struct sock *sk, unsigned char on, 775 void (*destructor)(struct sock *)); 776 777 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); 778 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 779 u32 info, u8 *payload); 780 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, 781 u32 info); 782 783 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 784 { 785 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0); 786 } 787 788 bool icmp_global_allow(void); 789 extern int sysctl_icmp_msgs_per_sec; 790 extern int sysctl_icmp_msgs_burst; 791 792 #ifdef CONFIG_PROC_FS 793 int ip_misc_proc_init(void); 794 #endif 795 796 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family, 797 struct netlink_ext_ack *extack); 798 799 static inline bool inetdev_valid_mtu(unsigned int mtu) 800 { 801 return likely(mtu >= IPV4_MIN_MTU); 802 } 803 804 void ip_sock_set_freebind(struct sock *sk); 805 int ip_sock_set_mtu_discover(struct sock *sk, int val); 806 void ip_sock_set_pktinfo(struct sock *sk); 807 void ip_sock_set_recverr(struct sock *sk); 808 void ip_sock_set_tos(struct sock *sk, int val); 809 void __ip_sock_set_tos(struct sock *sk, int val); 810 811 #endif /* _IP_H */ 812