ip.h 23 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835
  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. #include <linux/types.h>
  21. #include <linux/ip.h>
  22. #include <linux/in.h>
  23. #include <linux/skbuff.h>
  24. #include <linux/jhash.h>
  25. #include <linux/sockptr.h>
  26. #include <linux/static_key.h>
  27. #include <net/inet_sock.h>
  28. #include <net/route.h>
  29. #include <net/snmp.h>
  30. #include <net/flow.h>
  31. #include <net/flow_dissector.h>
  32. #include <net/netns/hash.h>
  33. #include <net/lwtunnel.h>
  34. #include <net/inet_dscp.h>
  35. #define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */
  36. #define IPV4_MIN_MTU 68 /* RFC 791 */
  37. extern unsigned int sysctl_fib_sync_mem;
  38. extern unsigned int sysctl_fib_sync_mem_min;
  39. extern unsigned int sysctl_fib_sync_mem_max;
  40. struct sock;
  41. struct inet_skb_parm {
  42. int iif;
  43. struct ip_options opt; /* Compiled IP options */
  44. u16 flags;
  45. #define IPSKB_FORWARDED BIT(0)
  46. #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
  47. #define IPSKB_XFRM_TRANSFORMED BIT(2)
  48. #define IPSKB_FRAG_COMPLETE BIT(3)
  49. #define IPSKB_REROUTED BIT(4)
  50. #define IPSKB_DOREDIRECT BIT(5)
  51. #define IPSKB_FRAG_PMTU BIT(6)
  52. #define IPSKB_L3SLAVE BIT(7)
  53. #define IPSKB_NOPOLICY BIT(8)
  54. #define IPSKB_MULTIPATH BIT(9)
  55. u16 frag_max_size;
  56. };
  57. static inline bool ipv4_l3mdev_skb(u16 flags)
  58. {
  59. return !!(flags & IPSKB_L3SLAVE);
  60. }
  61. static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
  62. {
  63. return ip_hdr(skb)->ihl * 4;
  64. }
  65. struct ipcm_cookie {
  66. struct sockcm_cookie sockc;
  67. __be32 addr;
  68. int oif;
  69. struct ip_options_rcu *opt;
  70. __u8 protocol;
  71. __u8 ttl;
  72. __s16 tos;
  73. char priority;
  74. __u16 gso_size;
  75. };
  76. static inline void ipcm_init(struct ipcm_cookie *ipcm)
  77. {
  78. *ipcm = (struct ipcm_cookie) { .tos = -1 };
  79. }
  80. static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
  81. const struct inet_sock *inet)
  82. {
  83. ipcm_init(ipcm);
  84. ipcm->sockc.mark = READ_ONCE(inet->sk.sk_mark);
  85. ipcm->sockc.tsflags = READ_ONCE(inet->sk.sk_tsflags);
  86. ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if);
  87. ipcm->addr = inet->inet_saddr;
  88. ipcm->protocol = inet->inet_num;
  89. }
  90. #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
  91. #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
  92. /* return enslaved device index if relevant */
  93. static inline int inet_sdif(const struct sk_buff *skb)
  94. {
  95. #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
  96. if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
  97. return IPCB(skb)->iif;
  98. #endif
  99. return 0;
  100. }
  101. /* Special input handler for packets caught by router alert option.
  102. They are selected only by protocol field, and then processed likely
  103. local ones; but only if someone wants them! Otherwise, router
  104. not running rsvpd will kill RSVP.
  105. It is user level problem, what it will make with them.
  106. I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
  107. but receiver should be enough clever f.e. to forward mtrace requests,
  108. sent to multicast group to reach destination designated router.
  109. */
  110. struct ip_ra_chain {
  111. struct ip_ra_chain __rcu *next;
  112. struct sock *sk;
  113. union {
  114. void (*destructor)(struct sock *);
  115. struct sock *saved_sk;
  116. };
  117. struct rcu_head rcu;
  118. };
  119. /* IP flags. */
  120. #define IP_CE 0x8000 /* Flag: "Congestion" */
  121. #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
  122. #define IP_MF 0x2000 /* Flag: "More Fragments" */
  123. #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
  124. #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
  125. struct msghdr;
  126. struct net_device;
  127. struct packet_type;
  128. struct rtable;
  129. struct sockaddr;
  130. int igmp_mc_init(void);
  131. /*
  132. * Functions provided by ip.c
  133. */
  134. int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
  135. __be32 saddr, __be32 daddr,
  136. struct ip_options_rcu *opt, u8 tos);
  137. int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
  138. struct net_device *orig_dev);
  139. void ip_list_rcv(struct list_head *head, struct packet_type *pt,
  140. struct net_device *orig_dev);
  141. int ip_local_deliver(struct sk_buff *skb);
  142. void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
  143. int ip_mr_input(struct sk_buff *skb);
  144. int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
  145. int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
  146. int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
  147. int (*output)(struct net *, struct sock *, struct sk_buff *));
  148. struct ip_fraglist_iter {
  149. struct sk_buff *frag;
  150. struct iphdr *iph;
  151. int offset;
  152. unsigned int hlen;
  153. };
  154. void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
  155. unsigned int hlen, struct ip_fraglist_iter *iter);
  156. void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
  157. static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
  158. {
  159. struct sk_buff *skb = iter->frag;
  160. iter->frag = skb->next;
  161. skb_mark_not_on_list(skb);
  162. return skb;
  163. }
  164. struct ip_frag_state {
  165. bool DF;
  166. unsigned int hlen;
  167. unsigned int ll_rs;
  168. unsigned int mtu;
  169. unsigned int left;
  170. int offset;
  171. int ptr;
  172. __be16 not_last_frag;
  173. };
  174. void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
  175. unsigned int mtu, bool DF, struct ip_frag_state *state);
  176. struct sk_buff *ip_frag_next(struct sk_buff *skb,
  177. struct ip_frag_state *state);
  178. void ip_send_check(struct iphdr *ip);
  179. int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
  180. int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
  181. int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
  182. __u8 tos);
  183. void ip_init(void);
  184. int ip_append_data(struct sock *sk, struct flowi4 *fl4,
  185. int getfrag(void *from, char *to, int offset, int len,
  186. int odd, struct sk_buff *skb),
  187. void *from, int len, int protolen,
  188. struct ipcm_cookie *ipc,
  189. struct rtable **rt,
  190. unsigned int flags);
  191. int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
  192. struct sk_buff *skb);
  193. struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
  194. struct sk_buff_head *queue,
  195. struct inet_cork *cork);
  196. int ip_send_skb(struct net *net, struct sk_buff *skb);
  197. int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
  198. void ip_flush_pending_frames(struct sock *sk);
  199. struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
  200. int getfrag(void *from, char *to, int offset,
  201. int len, int odd, struct sk_buff *skb),
  202. void *from, int length, int transhdrlen,
  203. struct ipcm_cookie *ipc, struct rtable **rtp,
  204. struct inet_cork *cork, unsigned int flags);
  205. int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
  206. static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
  207. {
  208. return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
  209. }
  210. /* Get the route scope that should be used when sending a packet. */
  211. static inline u8 ip_sendmsg_scope(const struct inet_sock *inet,
  212. const struct ipcm_cookie *ipc,
  213. const struct msghdr *msg)
  214. {
  215. if (sock_flag(&inet->sk, SOCK_LOCALROUTE) ||
  216. msg->msg_flags & MSG_DONTROUTE ||
  217. (ipc->opt && ipc->opt->opt.is_strictroute))
  218. return RT_SCOPE_LINK;
  219. return RT_SCOPE_UNIVERSE;
  220. }
  221. static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
  222. {
  223. u8 dsfield = ipc->tos != -1 ? ipc->tos : READ_ONCE(inet->tos);
  224. return dsfield & INET_DSCP_MASK;
  225. }
  226. /* datagram.c */
  227. int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
  228. int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
  229. void ip4_datagram_release_cb(struct sock *sk);
  230. struct ip_reply_arg {
  231. struct kvec iov[1];
  232. int flags;
  233. __wsum csum;
  234. int csumoffset; /* u16 offset of csum in iov[0].iov_base */
  235. /* -1 if not needed */
  236. int bound_dev_if;
  237. u8 tos;
  238. kuid_t uid;
  239. };
  240. #define IP_REPLY_ARG_NOSRCCHECK 1
  241. static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
  242. {
  243. return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
  244. }
  245. void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
  246. const struct ip_options *sopt,
  247. __be32 daddr, __be32 saddr,
  248. const struct ip_reply_arg *arg,
  249. unsigned int len, u64 transmit_time, u32 txhash);
  250. #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
  251. #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
  252. #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
  253. #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
  254. #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
  255. #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
  256. #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
  257. #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
  258. #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
  259. #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
  260. static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt)
  261. {
  262. return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt);
  263. }
  264. unsigned long snmp_fold_field(void __percpu *mib, int offt);
  265. #if BITS_PER_LONG==32
  266. u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
  267. size_t syncp_offset);
  268. u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
  269. #else
  270. static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
  271. size_t syncp_offset)
  272. {
  273. return snmp_get_cpu_field(mib, cpu, offct);
  274. }
  275. static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
  276. {
  277. return snmp_fold_field(mib, offt);
  278. }
  279. #endif
  280. #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
  281. { \
  282. int i, c; \
  283. for_each_possible_cpu(c) { \
  284. for (i = 0; stats_list[i].name; i++) \
  285. buff64[i] += snmp_get_cpu_field64( \
  286. mib_statistic, \
  287. c, stats_list[i].entry, \
  288. offset); \
  289. } \
  290. }
  291. #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
  292. { \
  293. int i, c; \
  294. for_each_possible_cpu(c) { \
  295. for (i = 0; stats_list[i].name; i++) \
  296. buff[i] += snmp_get_cpu_field( \
  297. mib_statistic, \
  298. c, stats_list[i].entry); \
  299. } \
  300. }
  301. static inline void inet_get_local_port_range(const struct net *net, int *low, int *high)
  302. {
  303. u32 range = READ_ONCE(net->ipv4.ip_local_ports.range);
  304. *low = range & 0xffff;
  305. *high = range >> 16;
  306. }
  307. bool inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high);
  308. #ifdef CONFIG_SYSCTL
  309. static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
  310. {
  311. if (!net->ipv4.sysctl_local_reserved_ports)
  312. return false;
  313. return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
  314. }
  315. static inline bool sysctl_dev_name_is_allowed(const char *name)
  316. {
  317. return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
  318. }
  319. static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
  320. {
  321. return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
  322. }
  323. #else
  324. static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
  325. {
  326. return false;
  327. }
  328. static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
  329. {
  330. return port < PROT_SOCK;
  331. }
  332. #endif
  333. __be32 inet_current_timestamp(void);
  334. /* From inetpeer.c */
  335. extern int inet_peer_threshold;
  336. extern int inet_peer_minttl;
  337. extern int inet_peer_maxttl;
  338. void ipfrag_init(void);
  339. void ip_static_sysctl_init(void);
  340. #define IP4_REPLY_MARK(net, mark) \
  341. (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
  342. static inline bool ip_is_fragment(const struct iphdr *iph)
  343. {
  344. return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
  345. }
  346. #ifdef CONFIG_INET
  347. #include <net/dst.h>
  348. /* The function in 2.2 was invalid, producing wrong result for
  349. * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
  350. static inline
  351. int ip_decrease_ttl(struct iphdr *iph)
  352. {
  353. u32 check = (__force u32)iph->check;
  354. check += (__force u32)htons(0x0100);
  355. iph->check = (__force __sum16)(check + (check>=0xFFFF));
  356. return --iph->ttl;
  357. }
  358. static inline dscp_t ip4h_dscp(const struct iphdr *ip4h)
  359. {
  360. return inet_dsfield_to_dscp(ip4h->tos);
  361. }
  362. static inline int ip_mtu_locked(const struct dst_entry *dst)
  363. {
  364. const struct rtable *rt = dst_rtable(dst);
  365. return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
  366. }
  367. static inline
  368. int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
  369. {
  370. u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
  371. return pmtudisc == IP_PMTUDISC_DO ||
  372. (pmtudisc == IP_PMTUDISC_WANT &&
  373. !ip_mtu_locked(dst));
  374. }
  375. static inline bool ip_sk_accept_pmtu(const struct sock *sk)
  376. {
  377. u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
  378. return pmtudisc != IP_PMTUDISC_INTERFACE &&
  379. pmtudisc != IP_PMTUDISC_OMIT;
  380. }
  381. static inline bool ip_sk_use_pmtu(const struct sock *sk)
  382. {
  383. return READ_ONCE(inet_sk(sk)->pmtudisc) < IP_PMTUDISC_PROBE;
  384. }
  385. static inline bool ip_sk_ignore_df(const struct sock *sk)
  386. {
  387. u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
  388. return pmtudisc < IP_PMTUDISC_DO || pmtudisc == IP_PMTUDISC_OMIT;
  389. }
  390. static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
  391. bool forwarding)
  392. {
  393. const struct rtable *rt = dst_rtable(dst);
  394. unsigned int mtu, res;
  395. struct net *net;
  396. rcu_read_lock();
  397. net = dev_net_rcu(dst_dev(dst));
  398. if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
  399. ip_mtu_locked(dst) ||
  400. !forwarding) {
  401. mtu = rt->rt_pmtu;
  402. if (mtu && time_before(jiffies, rt->dst.expires))
  403. goto out;
  404. }
  405. /* 'forwarding = true' case should always honour route mtu */
  406. mtu = dst_metric_raw(dst, RTAX_MTU);
  407. if (mtu)
  408. goto out;
  409. mtu = READ_ONCE(dst_dev(dst)->mtu);
  410. if (unlikely(ip_mtu_locked(dst))) {
  411. if (rt->rt_uses_gateway && mtu > 576)
  412. mtu = 576;
  413. }
  414. out:
  415. mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
  416. res = mtu - lwtunnel_headroom(dst->lwtstate, mtu);
  417. rcu_read_unlock();
  418. return res;
  419. }
  420. static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
  421. const struct sk_buff *skb)
  422. {
  423. const struct dst_entry *dst = skb_dst(skb);
  424. unsigned int mtu;
  425. if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
  426. bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
  427. return ip_dst_mtu_maybe_forward(dst, forwarding);
  428. }
  429. mtu = min(READ_ONCE(dst_dev(dst)->mtu), IP_MAX_MTU);
  430. return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
  431. }
  432. struct dst_metrics *ip_fib_metrics_init(struct nlattr *fc_mx, int fc_mx_len,
  433. struct netlink_ext_ack *extack);
  434. static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
  435. {
  436. if (fib_metrics != &dst_default_metrics &&
  437. refcount_dec_and_test(&fib_metrics->refcnt))
  438. kfree(fib_metrics);
  439. }
  440. /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
  441. static inline
  442. void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
  443. {
  444. dst_init_metrics(dst, fib_metrics->metrics, true);
  445. if (fib_metrics != &dst_default_metrics) {
  446. dst->_metrics |= DST_METRICS_REFCOUNTED;
  447. refcount_inc(&fib_metrics->refcnt);
  448. }
  449. }
  450. static inline
  451. void ip_dst_metrics_put(struct dst_entry *dst)
  452. {
  453. struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
  454. if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
  455. kfree(p);
  456. }
  457. void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
  458. static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
  459. struct sock *sk, int segs)
  460. {
  461. struct iphdr *iph = ip_hdr(skb);
  462. /* We had many attacks based on IPID, use the private
  463. * generator as much as we can.
  464. */
  465. if (sk && inet_sk(sk)->inet_daddr) {
  466. int val;
  467. /* avoid atomic operations for TCP,
  468. * as we hold socket lock at this point.
  469. */
  470. if (sk_is_tcp(sk)) {
  471. sock_owned_by_me(sk);
  472. val = atomic_read(&inet_sk(sk)->inet_id);
  473. atomic_set(&inet_sk(sk)->inet_id, val + segs);
  474. } else {
  475. val = atomic_add_return(segs, &inet_sk(sk)->inet_id);
  476. }
  477. iph->id = htons(val);
  478. return;
  479. }
  480. if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
  481. iph->id = 0;
  482. } else {
  483. /* Unfortunately we need the big hammer to get a suitable IPID */
  484. __ip_select_ident(net, iph, segs);
  485. }
  486. }
  487. static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
  488. struct sock *sk)
  489. {
  490. ip_select_ident_segs(net, skb, sk, 1);
  491. }
  492. static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
  493. {
  494. return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
  495. skb->len, proto, 0);
  496. }
  497. /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
  498. * Equivalent to : flow->v4addrs.src = iph->saddr;
  499. * flow->v4addrs.dst = iph->daddr;
  500. */
  501. static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
  502. const struct iphdr *iph)
  503. {
  504. BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
  505. offsetof(typeof(flow->addrs), v4addrs.src) +
  506. sizeof(flow->addrs.v4addrs.src));
  507. memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs));
  508. flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
  509. }
  510. /*
  511. * Map a multicast IP onto multicast MAC for type ethernet.
  512. */
  513. static inline void ip_eth_mc_map(__be32 naddr, char *buf)
  514. {
  515. __u32 addr=ntohl(naddr);
  516. buf[0]=0x01;
  517. buf[1]=0x00;
  518. buf[2]=0x5e;
  519. buf[5]=addr&0xFF;
  520. addr>>=8;
  521. buf[4]=addr&0xFF;
  522. addr>>=8;
  523. buf[3]=addr&0x7F;
  524. }
  525. /*
  526. * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
  527. * Leave P_Key as 0 to be filled in by driver.
  528. */
  529. static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
  530. {
  531. __u32 addr;
  532. unsigned char scope = broadcast[5] & 0xF;
  533. buf[0] = 0; /* Reserved */
  534. buf[1] = 0xff; /* Multicast QPN */
  535. buf[2] = 0xff;
  536. buf[3] = 0xff;
  537. addr = ntohl(naddr);
  538. buf[4] = 0xff;
  539. buf[5] = 0x10 | scope; /* scope from broadcast address */
  540. buf[6] = 0x40; /* IPv4 signature */
  541. buf[7] = 0x1b;
  542. buf[8] = broadcast[8]; /* P_Key */
  543. buf[9] = broadcast[9];
  544. buf[10] = 0;
  545. buf[11] = 0;
  546. buf[12] = 0;
  547. buf[13] = 0;
  548. buf[14] = 0;
  549. buf[15] = 0;
  550. buf[19] = addr & 0xff;
  551. addr >>= 8;
  552. buf[18] = addr & 0xff;
  553. addr >>= 8;
  554. buf[17] = addr & 0xff;
  555. addr >>= 8;
  556. buf[16] = addr & 0x0f;
  557. }
  558. static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
  559. {
  560. if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
  561. memcpy(buf, broadcast, 4);
  562. else
  563. memcpy(buf, &naddr, sizeof(naddr));
  564. }
  565. #if IS_ENABLED(CONFIG_IPV6)
  566. #include <linux/ipv6.h>
  567. #endif
  568. static __inline__ void inet_reset_saddr(struct sock *sk)
  569. {
  570. inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
  571. #if IS_ENABLED(CONFIG_IPV6)
  572. if (sk->sk_family == PF_INET6) {
  573. struct ipv6_pinfo *np = inet6_sk(sk);
  574. memset(&np->saddr, 0, sizeof(np->saddr));
  575. memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
  576. }
  577. #endif
  578. }
  579. #endif
  580. static inline unsigned int ipv4_addr_hash(__be32 ip)
  581. {
  582. return (__force unsigned int) ip;
  583. }
  584. static inline u32 ipv4_portaddr_hash(const struct net *net,
  585. __be32 saddr,
  586. unsigned int port)
  587. {
  588. return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
  589. }
  590. bool ip_call_ra_chain(struct sk_buff *skb);
  591. /*
  592. * Functions provided by ip_fragment.c
  593. */
  594. enum ip_defrag_users {
  595. IP_DEFRAG_LOCAL_DELIVER,
  596. IP_DEFRAG_CALL_RA_CHAIN,
  597. IP_DEFRAG_CONNTRACK_IN,
  598. __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
  599. IP_DEFRAG_CONNTRACK_OUT,
  600. __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
  601. IP_DEFRAG_CONNTRACK_BRIDGE_IN,
  602. __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
  603. IP_DEFRAG_VS_IN,
  604. IP_DEFRAG_VS_OUT,
  605. IP_DEFRAG_VS_FWD,
  606. IP_DEFRAG_AF_PACKET,
  607. IP_DEFRAG_MACVLAN,
  608. };
  609. /* Return true if the value of 'user' is between 'lower_bond'
  610. * and 'upper_bond' inclusively.
  611. */
  612. static inline bool ip_defrag_user_in_between(u32 user,
  613. enum ip_defrag_users lower_bond,
  614. enum ip_defrag_users upper_bond)
  615. {
  616. return user >= lower_bond && user <= upper_bond;
  617. }
  618. int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
  619. #ifdef CONFIG_INET
  620. struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
  621. #else
  622. static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
  623. {
  624. return skb;
  625. }
  626. #endif
  627. /*
  628. * Functions provided by ip_forward.c
  629. */
  630. int ip_forward(struct sk_buff *skb);
  631. /*
  632. * Functions provided by ip_options.c
  633. */
  634. void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
  635. __be32 daddr, struct rtable *rt);
  636. int __ip_options_echo(struct net *net, struct ip_options *dopt,
  637. struct sk_buff *skb, const struct ip_options *sopt);
  638. static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
  639. struct sk_buff *skb)
  640. {
  641. return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
  642. }
  643. void ip_options_fragment(struct sk_buff *skb);
  644. int __ip_options_compile(struct net *net, struct ip_options *opt,
  645. struct sk_buff *skb, __be32 *info);
  646. int ip_options_compile(struct net *net, struct ip_options *opt,
  647. struct sk_buff *skb);
  648. int ip_options_get(struct net *net, struct ip_options_rcu **optp,
  649. sockptr_t data, int optlen);
  650. void ip_options_undo(struct ip_options *opt);
  651. void ip_forward_options(struct sk_buff *skb);
  652. int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
  653. /*
  654. * Functions provided by ip_sockglue.c
  655. */
  656. void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb, bool drop_dst);
  657. void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
  658. struct sk_buff *skb, int tlen, int offset);
  659. int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
  660. struct ipcm_cookie *ipc, bool allow_ipv6);
  661. DECLARE_STATIC_KEY_FALSE(ip4_min_ttl);
  662. int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
  663. unsigned int optlen);
  664. int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
  665. unsigned int optlen);
  666. int do_ip_getsockopt(struct sock *sk, int level, int optname,
  667. sockptr_t optval, sockptr_t optlen);
  668. int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
  669. int __user *optlen);
  670. int ip_ra_control(struct sock *sk, unsigned char on,
  671. void (*destructor)(struct sock *));
  672. int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
  673. void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
  674. u32 info, u8 *payload);
  675. void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
  676. u32 info);
  677. static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
  678. {
  679. ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
  680. }
  681. bool icmp_global_allow(struct net *net);
  682. void icmp_global_consume(struct net *net);
  683. #ifdef CONFIG_PROC_FS
  684. int ip_misc_proc_init(void);
  685. #endif
  686. int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
  687. struct netlink_ext_ack *extack);
  688. static inline bool inetdev_valid_mtu(unsigned int mtu)
  689. {
  690. return likely(mtu >= IPV4_MIN_MTU);
  691. }
  692. void ip_sock_set_freebind(struct sock *sk);
  693. int ip_sock_set_mtu_discover(struct sock *sk, int val);
  694. void ip_sock_set_pktinfo(struct sock *sk);
  695. void ip_sock_set_recverr(struct sock *sk);
  696. void ip_sock_set_tos(struct sock *sk, int val);
  697. void __ip_sock_set_tos(struct sock *sk, int val);
  698. #endif /* _IP_H */