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 #include <net/inet_dscp.h> 37 38 #define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */ 39 #define IPV4_MIN_MTU 68 /* RFC 791 */ 40 41 extern unsigned int sysctl_fib_sync_mem; 42 extern unsigned int sysctl_fib_sync_mem_min; 43 extern unsigned int sysctl_fib_sync_mem_max; 44 45 struct sock; 46 47 struct inet_skb_parm { 48 int iif; 49 struct ip_options opt; /* Compiled IP options */ 50 u16 flags; 51 52 #define IPSKB_FORWARDED BIT(0) 53 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1) 54 #define IPSKB_XFRM_TRANSFORMED BIT(2) 55 #define IPSKB_FRAG_COMPLETE BIT(3) 56 #define IPSKB_REROUTED BIT(4) 57 #define IPSKB_DOREDIRECT BIT(5) 58 #define IPSKB_FRAG_PMTU BIT(6) 59 #define IPSKB_L3SLAVE BIT(7) 60 #define IPSKB_NOPOLICY BIT(8) 61 #define IPSKB_MULTIPATH BIT(9) 62 #define IPSKB_MCROUTE BIT(10) 63 64 u16 frag_max_size; 65 }; 66 67 static inline bool ipv4_l3mdev_skb(u16 flags) 68 { 69 return !!(flags & IPSKB_L3SLAVE); 70 } 71 72 static inline unsigned int ip_hdrlen(const struct sk_buff *skb) 73 { 74 return ip_hdr(skb)->ihl * 4; 75 } 76 77 struct ipcm_cookie { 78 struct sockcm_cookie sockc; 79 __be32 addr; 80 int oif; 81 struct ip_options_rcu *opt; 82 __u8 protocol; 83 __u8 ttl; 84 __s16 tos; 85 __u16 gso_size; 86 }; 87 88 static inline void ipcm_init(struct ipcm_cookie *ipcm) 89 { 90 *ipcm = (struct ipcm_cookie) { .tos = -1 }; 91 } 92 93 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm, 94 const struct inet_sock *inet) 95 { 96 *ipcm = (struct ipcm_cookie) { 97 .tos = READ_ONCE(inet->tos), 98 }; 99 100 sockcm_init(&ipcm->sockc, &inet->sk); 101 102 ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if); 103 ipcm->addr = inet->inet_saddr; 104 ipcm->protocol = inet->inet_num; 105 } 106 107 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) 108 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) 109 110 /* return enslaved device index if relevant */ 111 static inline int inet_sdif(const struct sk_buff *skb) 112 { 113 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV) 114 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags)) 115 return IPCB(skb)->iif; 116 #endif 117 return 0; 118 } 119 120 /* Special input handler for packets caught by router alert option. 121 They are selected only by protocol field, and then processed likely 122 local ones; but only if someone wants them! Otherwise, router 123 not running rsvpd will kill RSVP. 124 125 It is user level problem, what it will make with them. 126 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), 127 but receiver should be enough clever f.e. to forward mtrace requests, 128 sent to multicast group to reach destination designated router. 129 */ 130 131 struct ip_ra_chain { 132 struct ip_ra_chain __rcu *next; 133 struct sock *sk; 134 union { 135 void (*destructor)(struct sock *); 136 struct sock *saved_sk; 137 }; 138 struct rcu_head rcu; 139 }; 140 141 /* IP flags. */ 142 #define IP_CE 0x8000 /* Flag: "Congestion" */ 143 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */ 144 #define IP_MF 0x2000 /* Flag: "More Fragments" */ 145 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ 146 147 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ 148 149 struct msghdr; 150 struct net_device; 151 struct packet_type; 152 struct rtable; 153 struct sockaddr; 154 155 int igmp_mc_init(void); 156 157 /* 158 * Functions provided by ip.c 159 */ 160 161 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk, 162 __be32 saddr, __be32 daddr, 163 struct ip_options_rcu *opt, u8 tos); 164 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, 165 struct net_device *orig_dev); 166 void ip_list_rcv(struct list_head *head, struct packet_type *pt, 167 struct net_device *orig_dev); 168 int ip_local_deliver(struct sk_buff *skb); 169 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto); 170 int ip_mr_input(struct sk_buff *skb); 171 int ip_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb); 172 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb); 173 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb); 174 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, 175 int (*output)(struct net *, struct sock *, struct sk_buff *)); 176 177 struct ip_fraglist_iter { 178 struct sk_buff *frag; 179 struct iphdr *iph; 180 int offset; 181 unsigned int hlen; 182 }; 183 184 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph, 185 unsigned int hlen, struct ip_fraglist_iter *iter); 186 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter); 187 188 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter) 189 { 190 struct sk_buff *skb = iter->frag; 191 192 iter->frag = skb->next; 193 skb_mark_not_on_list(skb); 194 195 return skb; 196 } 197 198 struct ip_frag_state { 199 bool DF; 200 unsigned int hlen; 201 unsigned int ll_rs; 202 unsigned int mtu; 203 unsigned int left; 204 int offset; 205 int ptr; 206 __be16 not_last_frag; 207 }; 208 209 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs, 210 unsigned int mtu, bool DF, struct ip_frag_state *state); 211 struct sk_buff *ip_frag_next(struct sk_buff *skb, 212 struct ip_frag_state *state); 213 214 void ip_send_check(struct iphdr *ip); 215 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 216 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); 217 218 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl, 219 __u8 tos); 220 void ip_init(void); 221 int ip_append_data(struct sock *sk, struct flowi4 *fl4, 222 int getfrag(void *from, char *to, int offset, int len, 223 int odd, struct sk_buff *skb), 224 void *from, int len, int protolen, 225 struct ipcm_cookie *ipc, 226 struct rtable **rt, 227 unsigned int flags); 228 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, 229 struct sk_buff *skb); 230 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, 231 struct sk_buff_head *queue, 232 struct inet_cork *cork); 233 int ip_send_skb(struct net *net, struct sk_buff *skb); 234 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); 235 void ip_flush_pending_frames(struct sock *sk); 236 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, 237 int getfrag(void *from, char *to, int offset, 238 int len, int odd, struct sk_buff *skb), 239 void *from, int length, int transhdrlen, 240 struct ipcm_cookie *ipc, struct rtable **rtp, 241 struct inet_cork *cork, unsigned int flags); 242 243 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); 244 245 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) 246 { 247 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base); 248 } 249 250 /* Get the route scope that should be used when sending a packet. */ 251 static inline u8 ip_sendmsg_scope(const struct inet_sock *inet, 252 const struct ipcm_cookie *ipc, 253 const struct msghdr *msg) 254 { 255 if (sock_flag(&inet->sk, SOCK_LOCALROUTE) || 256 msg->msg_flags & MSG_DONTROUTE || 257 (ipc->opt && ipc->opt->opt.is_strictroute)) 258 return RT_SCOPE_LINK; 259 260 return RT_SCOPE_UNIVERSE; 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, const struct sock *orig_sk, 288 struct sk_buff *skb, 289 const struct ip_options *sopt, 290 __be32 daddr, __be32 saddr, 291 const struct ip_reply_arg *arg, 292 unsigned int len, u64 transmit_time, u32 txhash); 293 294 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) 295 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field) 296 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) 297 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_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 __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) 300 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) 301 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field) 302 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 303 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) 304 305 static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt) 306 { 307 return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt); 308 } 309 310 unsigned long snmp_fold_field(void __percpu *mib, int offt); 311 #if BITS_PER_LONG==32 312 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 313 size_t syncp_offset); 314 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); 315 #else 316 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, 317 size_t syncp_offset) 318 { 319 return snmp_get_cpu_field(mib, cpu, offct); 320 321 } 322 323 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) 324 { 325 return snmp_fold_field(mib, offt); 326 } 327 #endif 328 329 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \ 330 { \ 331 int i, c; \ 332 for_each_possible_cpu(c) { \ 333 for (i = 0; stats_list[i].name; i++) \ 334 buff64[i] += snmp_get_cpu_field64( \ 335 mib_statistic, \ 336 c, stats_list[i].entry, \ 337 offset); \ 338 } \ 339 } 340 341 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \ 342 { \ 343 int i, c; \ 344 for_each_possible_cpu(c) { \ 345 for (i = 0; stats_list[i].name; i++) \ 346 buff[i] += snmp_get_cpu_field( \ 347 mib_statistic, \ 348 c, stats_list[i].entry); \ 349 } \ 350 } 351 352 static inline void inet_get_local_port_range(const struct net *net, int *low, int *high) 353 { 354 u32 range = READ_ONCE(net->ipv4.ip_local_ports.range); 355 356 *low = range & 0xffff; 357 *high = range >> 16; 358 } 359 bool inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high); 360 361 #ifdef CONFIG_SYSCTL 362 static inline bool inet_is_local_reserved_port(const struct net *net, unsigned short port) 363 { 364 if (!net->ipv4.sysctl_local_reserved_ports) 365 return false; 366 return test_bit(port, net->ipv4.sysctl_local_reserved_ports); 367 } 368 369 static inline bool sysctl_dev_name_is_allowed(const char *name) 370 { 371 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; 372 } 373 374 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) 375 { 376 return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock); 377 } 378 379 #else 380 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) 381 { 382 return false; 383 } 384 385 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) 386 { 387 return port < PROT_SOCK; 388 } 389 #endif 390 391 __be32 inet_current_timestamp(void); 392 393 /* From inetpeer.c */ 394 extern int inet_peer_threshold; 395 extern int inet_peer_minttl; 396 extern int inet_peer_maxttl; 397 398 void ipfrag_init(void); 399 400 void ip_static_sysctl_init(void); 401 402 #define IP4_REPLY_MARK(net, mark) \ 403 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0) 404 405 static inline bool ip_is_fragment(const struct iphdr *iph) 406 { 407 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; 408 } 409 410 #ifdef CONFIG_INET 411 #include <net/dst.h> 412 413 /* The function in 2.2 was invalid, producing wrong result for 414 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ 415 static inline 416 int ip_decrease_ttl(struct iphdr *iph) 417 { 418 u32 check = (__force u32)iph->check; 419 check += (__force u32)htons(0x0100); 420 iph->check = (__force __sum16)(check + (check>=0xFFFF)); 421 return --iph->ttl; 422 } 423 424 static inline dscp_t ip4h_dscp(const struct iphdr *ip4h) 425 { 426 return inet_dsfield_to_dscp(ip4h->tos); 427 } 428 429 static inline int ip_mtu_locked(const struct dst_entry *dst) 430 { 431 const struct rtable *rt = dst_rtable(dst); 432 433 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU); 434 } 435 436 static inline 437 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst) 438 { 439 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); 440 441 return pmtudisc == IP_PMTUDISC_DO || 442 (pmtudisc == IP_PMTUDISC_WANT && 443 !ip_mtu_locked(dst)); 444 } 445 446 static inline bool ip_sk_accept_pmtu(const struct sock *sk) 447 { 448 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); 449 450 return pmtudisc != IP_PMTUDISC_INTERFACE && 451 pmtudisc != IP_PMTUDISC_OMIT; 452 } 453 454 static inline bool ip_sk_use_pmtu(const struct sock *sk) 455 { 456 return READ_ONCE(inet_sk(sk)->pmtudisc) < IP_PMTUDISC_PROBE; 457 } 458 459 static inline bool ip_sk_ignore_df(const struct sock *sk) 460 { 461 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); 462 463 return pmtudisc < IP_PMTUDISC_DO || pmtudisc == IP_PMTUDISC_OMIT; 464 } 465 466 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, 467 bool forwarding) 468 { 469 const struct rtable *rt = dst_rtable(dst); 470 unsigned int mtu, res; 471 struct net *net; 472 473 rcu_read_lock(); 474 475 net = dev_net_rcu(dst_dev(dst)); 476 if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) || 477 ip_mtu_locked(dst) || 478 !forwarding) { 479 mtu = rt->rt_pmtu; 480 if (mtu && time_before(jiffies, READ_ONCE(rt->dst.expires))) 481 goto out; 482 } 483 484 /* 'forwarding = true' case should always honour route mtu */ 485 mtu = dst_metric_raw(dst, RTAX_MTU); 486 if (mtu) 487 goto out; 488 489 mtu = READ_ONCE(dst_dev(dst)->mtu); 490 491 if (unlikely(ip_mtu_locked(dst))) { 492 if (rt->rt_uses_gateway && mtu > 576) 493 mtu = 576; 494 } 495 496 out: 497 mtu = min_t(unsigned int, mtu, IP_MAX_MTU); 498 499 res = mtu - lwtunnel_headroom(dst->lwtstate, mtu); 500 501 rcu_read_unlock(); 502 503 return res; 504 } 505 506 static inline unsigned int ip_skb_dst_mtu(struct sock *sk, 507 const struct sk_buff *skb) 508 { 509 const struct dst_entry *dst = skb_dst(skb); 510 unsigned int mtu; 511 512 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) { 513 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; 514 515 return ip_dst_mtu_maybe_forward(dst, forwarding); 516 } 517 518 mtu = min(READ_ONCE(dst_dev(dst)->mtu), IP_MAX_MTU); 519 return mtu - lwtunnel_headroom(dst->lwtstate, mtu); 520 } 521 522 struct dst_metrics *ip_fib_metrics_init(struct nlattr *fc_mx, int fc_mx_len, 523 struct netlink_ext_ack *extack); 524 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics) 525 { 526 if (fib_metrics != &dst_default_metrics && 527 refcount_dec_and_test(&fib_metrics->refcnt)) 528 kfree(fib_metrics); 529 } 530 531 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */ 532 static inline 533 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics) 534 { 535 dst_init_metrics(dst, fib_metrics->metrics, true); 536 537 if (fib_metrics != &dst_default_metrics) { 538 dst->_metrics |= DST_METRICS_REFCOUNTED; 539 refcount_inc(&fib_metrics->refcnt); 540 } 541 } 542 543 static inline 544 void ip_dst_metrics_put(struct dst_entry *dst) 545 { 546 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst); 547 548 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt)) 549 kfree(p); 550 } 551 552 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); 553 554 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, 555 struct sock *sk, int segs) 556 { 557 struct iphdr *iph = ip_hdr(skb); 558 559 /* We had many attacks based on IPID, use the private 560 * generator as much as we can. 561 */ 562 if (sk && inet_sk(sk)->inet_daddr) { 563 int val; 564 565 /* avoid atomic operations for TCP, 566 * as we hold socket lock at this point. 567 */ 568 if (sk_is_tcp(sk)) { 569 sock_owned_by_me(sk); 570 val = atomic_read(&inet_sk(sk)->inet_id); 571 atomic_set(&inet_sk(sk)->inet_id, val + segs); 572 } else { 573 val = atomic_add_return(segs, &inet_sk(sk)->inet_id); 574 } 575 iph->id = htons(val); 576 return; 577 } 578 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { 579 iph->id = 0; 580 } else { 581 /* Unfortunately we need the big hammer to get a suitable IPID */ 582 __ip_select_ident(net, iph, segs); 583 } 584 } 585 586 static inline void ip_select_ident(struct net *net, struct sk_buff *skb, 587 struct sock *sk) 588 { 589 ip_select_ident_segs(net, skb, sk, 1); 590 } 591 592 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) 593 { 594 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr, 595 skb->len, proto, 0); 596 } 597 598 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store 599 * Equivalent to : flow->v4addrs.src = iph->saddr; 600 * flow->v4addrs.dst = iph->daddr; 601 */ 602 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, 603 const struct iphdr *iph) 604 { 605 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != 606 offsetof(typeof(flow->addrs), v4addrs.src) + 607 sizeof(flow->addrs.v4addrs.src)); 608 memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs)); 609 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; 610 } 611 612 /* 613 * Map a multicast IP onto multicast MAC for type ethernet. 614 */ 615 616 static inline void ip_eth_mc_map(__be32 naddr, char *buf) 617 { 618 __u32 addr=ntohl(naddr); 619 buf[0]=0x01; 620 buf[1]=0x00; 621 buf[2]=0x5e; 622 buf[5]=addr&0xFF; 623 addr>>=8; 624 buf[4]=addr&0xFF; 625 addr>>=8; 626 buf[3]=addr&0x7F; 627 } 628 629 /* 630 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. 631 * Leave P_Key as 0 to be filled in by driver. 632 */ 633 634 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 635 { 636 __u32 addr; 637 unsigned char scope = broadcast[5] & 0xF; 638 639 buf[0] = 0; /* Reserved */ 640 buf[1] = 0xff; /* Multicast QPN */ 641 buf[2] = 0xff; 642 buf[3] = 0xff; 643 addr = ntohl(naddr); 644 buf[4] = 0xff; 645 buf[5] = 0x10 | scope; /* scope from broadcast address */ 646 buf[6] = 0x40; /* IPv4 signature */ 647 buf[7] = 0x1b; 648 buf[8] = broadcast[8]; /* P_Key */ 649 buf[9] = broadcast[9]; 650 buf[10] = 0; 651 buf[11] = 0; 652 buf[12] = 0; 653 buf[13] = 0; 654 buf[14] = 0; 655 buf[15] = 0; 656 buf[19] = addr & 0xff; 657 addr >>= 8; 658 buf[18] = addr & 0xff; 659 addr >>= 8; 660 buf[17] = addr & 0xff; 661 addr >>= 8; 662 buf[16] = addr & 0x0f; 663 } 664 665 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) 666 { 667 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) 668 memcpy(buf, broadcast, 4); 669 else 670 memcpy(buf, &naddr, sizeof(naddr)); 671 } 672 673 #if IS_ENABLED(CONFIG_IPV6) 674 #include <linux/ipv6.h> 675 #endif 676 677 static __inline__ void inet_reset_saddr(struct sock *sk) 678 { 679 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; 680 #if IS_ENABLED(CONFIG_IPV6) 681 if (sk->sk_family == PF_INET6) { 682 struct ipv6_pinfo *np = inet6_sk(sk); 683 684 memset(&np->saddr, 0, sizeof(np->saddr)); 685 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); 686 } 687 #endif 688 } 689 690 #endif 691 692 #if IS_MODULE(CONFIG_IPV6) 693 #define EXPORT_IPV6_MOD(X) EXPORT_SYMBOL(X) 694 #define EXPORT_IPV6_MOD_GPL(X) EXPORT_SYMBOL_GPL(X) 695 #else 696 #define EXPORT_IPV6_MOD(X) 697 #define EXPORT_IPV6_MOD_GPL(X) 698 #endif 699 700 static inline unsigned int ipv4_addr_hash(__be32 ip) 701 { 702 return (__force unsigned int) ip; 703 } 704 705 static inline u32 __ipv4_addr_hash(const __be32 ip, const u32 initval) 706 { 707 return jhash_1word((__force u32)ip, initval); 708 } 709 710 static inline u32 ipv4_portaddr_hash(const struct net *net, 711 __be32 saddr, 712 unsigned int port) 713 { 714 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port; 715 } 716 717 bool ip_call_ra_chain(struct sk_buff *skb); 718 719 /* 720 * Functions provided by ip_fragment.c 721 */ 722 723 enum ip_defrag_users { 724 IP_DEFRAG_LOCAL_DELIVER, 725 IP_DEFRAG_CALL_RA_CHAIN, 726 IP_DEFRAG_CONNTRACK_IN, 727 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, 728 IP_DEFRAG_CONNTRACK_OUT, 729 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, 730 IP_DEFRAG_CONNTRACK_BRIDGE_IN, 731 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, 732 IP_DEFRAG_VS_IN, 733 IP_DEFRAG_VS_OUT, 734 IP_DEFRAG_VS_FWD, 735 IP_DEFRAG_AF_PACKET, 736 IP_DEFRAG_MACVLAN, 737 }; 738 739 /* Return true if the value of 'user' is between 'lower_bond' 740 * and 'upper_bond' inclusively. 741 */ 742 static inline bool ip_defrag_user_in_between(u32 user, 743 enum ip_defrag_users lower_bond, 744 enum ip_defrag_users upper_bond) 745 { 746 return user >= lower_bond && user <= upper_bond; 747 } 748 749 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user); 750 #ifdef CONFIG_INET 751 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user); 752 #else 753 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user) 754 { 755 return skb; 756 } 757 #endif 758 759 /* 760 * Functions provided by ip_forward.c 761 */ 762 763 int ip_forward(struct sk_buff *skb); 764 765 /* 766 * Functions provided by ip_options.c 767 */ 768 769 void ip_options_build(struct sk_buff *skb, struct ip_options *opt, 770 __be32 daddr, struct rtable *rt); 771 772 int __ip_options_echo(struct net *net, struct ip_options *dopt, 773 struct sk_buff *skb, const struct ip_options *sopt); 774 static inline int ip_options_echo(struct net *net, struct ip_options *dopt, 775 struct sk_buff *skb) 776 { 777 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt); 778 } 779 780 void ip_options_fragment(struct sk_buff *skb); 781 int __ip_options_compile(struct net *net, struct ip_options *opt, 782 struct sk_buff *skb, __be32 *info); 783 int ip_options_compile(struct net *net, struct ip_options *opt, 784 struct sk_buff *skb); 785 int ip_options_get(struct net *net, struct ip_options_rcu **optp, 786 sockptr_t data, int optlen); 787 void ip_options_undo(struct ip_options *opt); 788 void ip_forward_options(struct sk_buff *skb); 789 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev); 790 791 /* 792 * Functions provided by ip_sockglue.c 793 */ 794 795 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb, bool drop_dst); 796 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk, 797 struct sk_buff *skb, int tlen, int offset); 798 int ip_cmsg_send(struct sock *sk, struct msghdr *msg, 799 struct ipcm_cookie *ipc, bool allow_ipv6); 800 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl); 801 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, 802 unsigned int optlen); 803 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, 804 unsigned int optlen); 805 int do_ip_getsockopt(struct sock *sk, int level, int optname, 806 sockptr_t optval, sockptr_t optlen); 807 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, 808 int __user *optlen); 809 int ip_ra_control(struct sock *sk, unsigned char on, 810 void (*destructor)(struct sock *)); 811 812 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); 813 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, 814 u32 info, u8 *payload); 815 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, 816 u32 info); 817 818 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) 819 { 820 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0); 821 } 822 823 bool icmp_global_allow(struct net *net); 824 void icmp_global_consume(struct net *net); 825 826 #ifdef CONFIG_PROC_FS 827 int ip_misc_proc_init(void); 828 #endif 829 830 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family, 831 struct netlink_ext_ack *extack); 832 833 static inline bool inetdev_valid_mtu(unsigned int mtu) 834 { 835 return likely(mtu >= IPV4_MIN_MTU); 836 } 837 838 void ip_sock_set_freebind(struct sock *sk); 839 int ip_sock_set_mtu_discover(struct sock *sk, int val); 840 void ip_sock_set_pktinfo(struct sock *sk); 841 void ip_sock_set_recverr(struct sock *sk); 842 void ip_sock_set_tos(struct sock *sk, int val); 843 void __ip_sock_set_tos(struct sock *sk, int val); 844 845 #endif /* _IP_H */ 846