blob: 52abfc00b5e3d3154cfd1b779ed81a00aef8d3c8 [file] [log] [blame]
David Brazdil0f672f62019-12-10 10:32:29 +00001/* SPDX-License-Identifier: GPL-2.0-or-later */
Andrew Scullb4b6d4a2019-01-02 15:54:55 +00002/*
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
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000017 */
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
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>
Olivier Deprez0e641232021-09-23 10:07:05 +020033#include <net/lwtunnel.h>
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000034
35#define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */
36#define IPV4_MIN_MTU 68 /* RFC 791 */
37
David Brazdil0f672f62019-12-10 10:32:29 +000038extern unsigned int sysctl_fib_sync_mem;
39extern unsigned int sysctl_fib_sync_mem_min;
40extern unsigned int sysctl_fib_sync_mem_max;
41
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000042struct sock;
43
44struct inet_skb_parm {
45 int iif;
46 struct ip_options opt; /* Compiled IP options */
47 u16 flags;
48
49#define IPSKB_FORWARDED BIT(0)
50#define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
51#define IPSKB_XFRM_TRANSFORMED BIT(2)
52#define IPSKB_FRAG_COMPLETE BIT(3)
53#define IPSKB_REROUTED BIT(4)
54#define IPSKB_DOREDIRECT BIT(5)
55#define IPSKB_FRAG_PMTU BIT(6)
56#define IPSKB_L3SLAVE BIT(7)
57
58 u16 frag_max_size;
59};
60
61static inline bool ipv4_l3mdev_skb(u16 flags)
62{
63 return !!(flags & IPSKB_L3SLAVE);
64}
65
66static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
67{
68 return ip_hdr(skb)->ihl * 4;
69}
70
71struct ipcm_cookie {
72 struct sockcm_cookie sockc;
73 __be32 addr;
74 int oif;
75 struct ip_options_rcu *opt;
76 __u8 ttl;
77 __s16 tos;
78 char priority;
79 __u16 gso_size;
80};
81
82static inline void ipcm_init(struct ipcm_cookie *ipcm)
83{
84 *ipcm = (struct ipcm_cookie) { .tos = -1 };
85}
86
87static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
88 const struct inet_sock *inet)
89{
90 ipcm_init(ipcm);
91
David Brazdil0f672f62019-12-10 10:32:29 +000092 ipcm->sockc.mark = inet->sk.sk_mark;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000093 ipcm->sockc.tsflags = inet->sk.sk_tsflags;
94 ipcm->oif = inet->sk.sk_bound_dev_if;
95 ipcm->addr = inet->inet_saddr;
96}
97
98#define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
99#define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
100
101/* return enslaved device index if relevant */
102static inline int inet_sdif(struct sk_buff *skb)
103{
104#if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
105 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
106 return IPCB(skb)->iif;
107#endif
108 return 0;
109}
110
111/* Special input handler for packets caught by router alert option.
112 They are selected only by protocol field, and then processed likely
113 local ones; but only if someone wants them! Otherwise, router
114 not running rsvpd will kill RSVP.
115
116 It is user level problem, what it will make with them.
117 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
118 but receiver should be enough clever f.e. to forward mtrace requests,
119 sent to multicast group to reach destination designated router.
120 */
121
122struct ip_ra_chain {
123 struct ip_ra_chain __rcu *next;
124 struct sock *sk;
125 union {
126 void (*destructor)(struct sock *);
127 struct sock *saved_sk;
128 };
129 struct rcu_head rcu;
130};
131
132/* IP flags. */
133#define IP_CE 0x8000 /* Flag: "Congestion" */
134#define IP_DF 0x4000 /* Flag: "Don't Fragment" */
135#define IP_MF 0x2000 /* Flag: "More Fragments" */
136#define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
137
138#define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
139
140struct msghdr;
141struct net_device;
142struct packet_type;
143struct rtable;
144struct sockaddr;
145
146int igmp_mc_init(void);
147
148/*
149 * Functions provided by ip.c
150 */
151
152int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
153 __be32 saddr, __be32 daddr,
154 struct ip_options_rcu *opt);
155int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
156 struct net_device *orig_dev);
157void ip_list_rcv(struct list_head *head, struct packet_type *pt,
158 struct net_device *orig_dev);
159int ip_local_deliver(struct sk_buff *skb);
David Brazdil0f672f62019-12-10 10:32:29 +0000160void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000161int ip_mr_input(struct sk_buff *skb);
162int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
163int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
164int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
165 int (*output)(struct net *, struct sock *, struct sk_buff *));
David Brazdil0f672f62019-12-10 10:32:29 +0000166
167struct ip_fraglist_iter {
168 struct sk_buff *frag;
169 struct iphdr *iph;
170 int offset;
171 unsigned int hlen;
172};
173
174void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
175 unsigned int hlen, struct ip_fraglist_iter *iter);
176void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
177
178static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
179{
180 struct sk_buff *skb = iter->frag;
181
182 iter->frag = skb->next;
183 skb_mark_not_on_list(skb);
184
185 return skb;
186}
187
188struct ip_frag_state {
189 bool DF;
190 unsigned int hlen;
191 unsigned int ll_rs;
192 unsigned int mtu;
193 unsigned int left;
194 int offset;
195 int ptr;
196 __be16 not_last_frag;
197};
198
199void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
200 unsigned int mtu, bool DF, struct ip_frag_state *state);
201struct sk_buff *ip_frag_next(struct sk_buff *skb,
202 struct ip_frag_state *state);
203
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000204void ip_send_check(struct iphdr *ip);
205int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
206int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
207
208int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
209 __u8 tos);
210void ip_init(void);
211int ip_append_data(struct sock *sk, struct flowi4 *fl4,
212 int getfrag(void *from, char *to, int offset, int len,
213 int odd, struct sk_buff *skb),
214 void *from, int len, int protolen,
215 struct ipcm_cookie *ipc,
216 struct rtable **rt,
217 unsigned int flags);
218int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
219 struct sk_buff *skb);
220ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
221 int offset, size_t size, int flags);
222struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
223 struct sk_buff_head *queue,
224 struct inet_cork *cork);
225int ip_send_skb(struct net *net, struct sk_buff *skb);
226int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
227void ip_flush_pending_frames(struct sock *sk);
228struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
229 int getfrag(void *from, char *to, int offset,
230 int len, int odd, struct sk_buff *skb),
231 void *from, int length, int transhdrlen,
232 struct ipcm_cookie *ipc, struct rtable **rtp,
233 struct inet_cork *cork, unsigned int flags);
234
235static inline int ip_queue_xmit(struct sock *sk, struct sk_buff *skb,
236 struct flowi *fl)
237{
238 return __ip_queue_xmit(sk, skb, fl, inet_sk(sk)->tos);
239}
240
241static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
242{
243 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
244}
245
246static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
247{
248 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
249}
250
251static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
252{
253 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
254}
255
256/* datagram.c */
257int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
258int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
259
260void ip4_datagram_release_cb(struct sock *sk);
261
262struct ip_reply_arg {
263 struct kvec iov[1];
264 int flags;
265 __wsum csum;
266 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
267 /* -1 if not needed */
268 int bound_dev_if;
269 u8 tos;
270 kuid_t uid;
271};
272
273#define IP_REPLY_ARG_NOSRCCHECK 1
274
275static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
276{
277 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
278}
279
280void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
281 const struct ip_options *sopt,
282 __be32 daddr, __be32 saddr,
283 const struct ip_reply_arg *arg,
David Brazdil0f672f62019-12-10 10:32:29 +0000284 unsigned int len, u64 transmit_time);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000285
286#define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
287#define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
288#define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
289#define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
290#define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
291#define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
292#define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
293#define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
294#define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
295#define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
296
297u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
298unsigned long snmp_fold_field(void __percpu *mib, int offt);
299#if BITS_PER_LONG==32
300u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
301 size_t syncp_offset);
302u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
303#else
304static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
305 size_t syncp_offset)
306{
307 return snmp_get_cpu_field(mib, cpu, offct);
308
309}
310
311static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
312{
313 return snmp_fold_field(mib, offt);
314}
315#endif
316
317#define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
318{ \
319 int i, c; \
320 for_each_possible_cpu(c) { \
321 for (i = 0; stats_list[i].name; i++) \
322 buff64[i] += snmp_get_cpu_field64( \
323 mib_statistic, \
324 c, stats_list[i].entry, \
325 offset); \
326 } \
327}
328
329#define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
330{ \
331 int i, c; \
332 for_each_possible_cpu(c) { \
333 for (i = 0; stats_list[i].name; i++) \
334 buff[i] += snmp_get_cpu_field( \
335 mib_statistic, \
336 c, stats_list[i].entry); \
337 } \
338}
339
340void inet_get_local_port_range(struct net *net, int *low, int *high);
341
342#ifdef CONFIG_SYSCTL
343static inline int inet_is_local_reserved_port(struct net *net, int port)
344{
345 if (!net->ipv4.sysctl_local_reserved_ports)
346 return 0;
347 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
348}
349
350static inline bool sysctl_dev_name_is_allowed(const char *name)
351{
352 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
353}
354
355static inline int inet_prot_sock(struct net *net)
356{
357 return net->ipv4.sysctl_ip_prot_sock;
358}
359
360#else
361static inline int inet_is_local_reserved_port(struct net *net, int port)
362{
363 return 0;
364}
365
366static inline int inet_prot_sock(struct net *net)
367{
368 return PROT_SOCK;
369}
370#endif
371
372__be32 inet_current_timestamp(void);
373
374/* From inetpeer.c */
375extern int inet_peer_threshold;
376extern int inet_peer_minttl;
377extern int inet_peer_maxttl;
378
379void ipfrag_init(void);
380
381void ip_static_sysctl_init(void);
382
383#define IP4_REPLY_MARK(net, mark) \
384 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
385
386static inline bool ip_is_fragment(const struct iphdr *iph)
387{
388 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
389}
390
391#ifdef CONFIG_INET
392#include <net/dst.h>
393
394/* The function in 2.2 was invalid, producing wrong result for
395 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
396static inline
397int ip_decrease_ttl(struct iphdr *iph)
398{
399 u32 check = (__force u32)iph->check;
400 check += (__force u32)htons(0x0100);
401 iph->check = (__force __sum16)(check + (check>=0xFFFF));
402 return --iph->ttl;
403}
404
405static inline int ip_mtu_locked(const struct dst_entry *dst)
406{
407 const struct rtable *rt = (const struct rtable *)dst;
408
409 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
410}
411
412static inline
413int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
414{
415 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
416
417 return pmtudisc == IP_PMTUDISC_DO ||
418 (pmtudisc == IP_PMTUDISC_WANT &&
419 !ip_mtu_locked(dst));
420}
421
422static inline bool ip_sk_accept_pmtu(const struct sock *sk)
423{
424 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
425 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
426}
427
428static inline bool ip_sk_use_pmtu(const struct sock *sk)
429{
430 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
431}
432
433static inline bool ip_sk_ignore_df(const struct sock *sk)
434{
435 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
436 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
437}
438
439static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
440 bool forwarding)
441{
442 struct net *net = dev_net(dst->dev);
Olivier Deprez0e641232021-09-23 10:07:05 +0200443 unsigned int mtu;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000444
445 if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
446 ip_mtu_locked(dst) ||
447 !forwarding)
448 return dst_mtu(dst);
449
Olivier Deprez0e641232021-09-23 10:07:05 +0200450 /* 'forwarding = true' case should always honour route mtu */
451 mtu = dst_metric_raw(dst, RTAX_MTU);
452 if (!mtu)
453 mtu = min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU);
454
455 return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000456}
457
458static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
459 const struct sk_buff *skb)
460{
Olivier Deprez0e641232021-09-23 10:07:05 +0200461 unsigned int mtu;
462
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000463 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
464 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
465
466 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
467 }
468
Olivier Deprez0e641232021-09-23 10:07:05 +0200469 mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
470 return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000471}
472
David Brazdil0f672f62019-12-10 10:32:29 +0000473struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx,
474 int fc_mx_len,
475 struct netlink_ext_ack *extack);
476static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
477{
478 if (fib_metrics != &dst_default_metrics &&
479 refcount_dec_and_test(&fib_metrics->refcnt))
480 kfree(fib_metrics);
481}
482
483/* ipv4 and ipv6 both use refcounted metrics if it is not the default */
484static inline
485void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
486{
487 dst_init_metrics(dst, fib_metrics->metrics, true);
488
489 if (fib_metrics != &dst_default_metrics) {
490 dst->_metrics |= DST_METRICS_REFCOUNTED;
491 refcount_inc(&fib_metrics->refcnt);
492 }
493}
494
495static inline
496void ip_dst_metrics_put(struct dst_entry *dst)
497{
498 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
499
500 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
501 kfree(p);
502}
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000503
504u32 ip_idents_reserve(u32 hash, int segs);
505void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
506
507static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
508 struct sock *sk, int segs)
509{
510 struct iphdr *iph = ip_hdr(skb);
511
512 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
513 /* This is only to work around buggy Windows95/2000
514 * VJ compression implementations. If the ID field
515 * does not change, they drop every other packet in
516 * a TCP stream using header compression.
517 */
518 if (sk && inet_sk(sk)->inet_daddr) {
519 iph->id = htons(inet_sk(sk)->inet_id);
520 inet_sk(sk)->inet_id += segs;
521 } else {
522 iph->id = 0;
523 }
524 } else {
525 __ip_select_ident(net, iph, segs);
526 }
527}
528
529static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
530 struct sock *sk)
531{
532 ip_select_ident_segs(net, skb, sk, 1);
533}
534
535static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
536{
537 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
538 skb->len, proto, 0);
539}
540
541/* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
542 * Equivalent to : flow->v4addrs.src = iph->saddr;
543 * flow->v4addrs.dst = iph->daddr;
544 */
545static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
546 const struct iphdr *iph)
547{
548 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
549 offsetof(typeof(flow->addrs), v4addrs.src) +
550 sizeof(flow->addrs.v4addrs.src));
551 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs));
552 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
553}
554
555static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
556{
557 const struct iphdr *iph = skb_gro_network_header(skb);
558
559 return csum_tcpudp_nofold(iph->saddr, iph->daddr,
560 skb_gro_len(skb), proto, 0);
561}
562
563/*
564 * Map a multicast IP onto multicast MAC for type ethernet.
565 */
566
567static inline void ip_eth_mc_map(__be32 naddr, char *buf)
568{
569 __u32 addr=ntohl(naddr);
570 buf[0]=0x01;
571 buf[1]=0x00;
572 buf[2]=0x5e;
573 buf[5]=addr&0xFF;
574 addr>>=8;
575 buf[4]=addr&0xFF;
576 addr>>=8;
577 buf[3]=addr&0x7F;
578}
579
580/*
581 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
582 * Leave P_Key as 0 to be filled in by driver.
583 */
584
585static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
586{
587 __u32 addr;
588 unsigned char scope = broadcast[5] & 0xF;
589
590 buf[0] = 0; /* Reserved */
591 buf[1] = 0xff; /* Multicast QPN */
592 buf[2] = 0xff;
593 buf[3] = 0xff;
594 addr = ntohl(naddr);
595 buf[4] = 0xff;
596 buf[5] = 0x10 | scope; /* scope from broadcast address */
597 buf[6] = 0x40; /* IPv4 signature */
598 buf[7] = 0x1b;
599 buf[8] = broadcast[8]; /* P_Key */
600 buf[9] = broadcast[9];
601 buf[10] = 0;
602 buf[11] = 0;
603 buf[12] = 0;
604 buf[13] = 0;
605 buf[14] = 0;
606 buf[15] = 0;
607 buf[19] = addr & 0xff;
608 addr >>= 8;
609 buf[18] = addr & 0xff;
610 addr >>= 8;
611 buf[17] = addr & 0xff;
612 addr >>= 8;
613 buf[16] = addr & 0x0f;
614}
615
616static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
617{
618 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
619 memcpy(buf, broadcast, 4);
620 else
621 memcpy(buf, &naddr, sizeof(naddr));
622}
623
624#if IS_ENABLED(CONFIG_IPV6)
625#include <linux/ipv6.h>
626#endif
627
628static __inline__ void inet_reset_saddr(struct sock *sk)
629{
630 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
631#if IS_ENABLED(CONFIG_IPV6)
632 if (sk->sk_family == PF_INET6) {
633 struct ipv6_pinfo *np = inet6_sk(sk);
634
635 memset(&np->saddr, 0, sizeof(np->saddr));
636 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
637 }
638#endif
639}
640
641#endif
642
643static inline unsigned int ipv4_addr_hash(__be32 ip)
644{
645 return (__force unsigned int) ip;
646}
647
648static inline u32 ipv4_portaddr_hash(const struct net *net,
649 __be32 saddr,
650 unsigned int port)
651{
652 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
653}
654
655bool ip_call_ra_chain(struct sk_buff *skb);
656
657/*
658 * Functions provided by ip_fragment.c
659 */
660
661enum ip_defrag_users {
662 IP_DEFRAG_LOCAL_DELIVER,
663 IP_DEFRAG_CALL_RA_CHAIN,
664 IP_DEFRAG_CONNTRACK_IN,
665 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
666 IP_DEFRAG_CONNTRACK_OUT,
667 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
668 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
669 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
670 IP_DEFRAG_VS_IN,
671 IP_DEFRAG_VS_OUT,
672 IP_DEFRAG_VS_FWD,
673 IP_DEFRAG_AF_PACKET,
674 IP_DEFRAG_MACVLAN,
675};
676
677/* Return true if the value of 'user' is between 'lower_bond'
678 * and 'upper_bond' inclusively.
679 */
680static inline bool ip_defrag_user_in_between(u32 user,
681 enum ip_defrag_users lower_bond,
682 enum ip_defrag_users upper_bond)
683{
684 return user >= lower_bond && user <= upper_bond;
685}
686
687int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
688#ifdef CONFIG_INET
689struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
690#else
691static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
692{
693 return skb;
694}
695#endif
696
697/*
698 * Functions provided by ip_forward.c
699 */
700
701int ip_forward(struct sk_buff *skb);
702
703/*
704 * Functions provided by ip_options.c
705 */
706
707void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
708 __be32 daddr, struct rtable *rt, int is_frag);
709
710int __ip_options_echo(struct net *net, struct ip_options *dopt,
711 struct sk_buff *skb, const struct ip_options *sopt);
712static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
713 struct sk_buff *skb)
714{
715 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
716}
717
718void ip_options_fragment(struct sk_buff *skb);
David Brazdil0f672f62019-12-10 10:32:29 +0000719int __ip_options_compile(struct net *net, struct ip_options *opt,
720 struct sk_buff *skb, __be32 *info);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000721int ip_options_compile(struct net *net, struct ip_options *opt,
722 struct sk_buff *skb);
723int ip_options_get(struct net *net, struct ip_options_rcu **optp,
724 unsigned char *data, int optlen);
725int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
726 unsigned char __user *data, int optlen);
727void ip_options_undo(struct ip_options *opt);
728void ip_forward_options(struct sk_buff *skb);
David Brazdil0f672f62019-12-10 10:32:29 +0000729int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000730
731/*
732 * Functions provided by ip_sockglue.c
733 */
734
735void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
736void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
737 struct sk_buff *skb, int tlen, int offset);
738int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
739 struct ipcm_cookie *ipc, bool allow_ipv6);
740int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
741 unsigned int optlen);
742int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
743 int __user *optlen);
744int compat_ip_setsockopt(struct sock *sk, int level, int optname,
745 char __user *optval, unsigned int optlen);
746int compat_ip_getsockopt(struct sock *sk, int level, int optname,
747 char __user *optval, int __user *optlen);
748int ip_ra_control(struct sock *sk, unsigned char on,
749 void (*destructor)(struct sock *));
750
751int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
752void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
753 u32 info, u8 *payload);
754void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
755 u32 info);
756
757static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
758{
759 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
760}
761
762bool icmp_global_allow(void);
763extern int sysctl_icmp_msgs_per_sec;
764extern int sysctl_icmp_msgs_burst;
765
766#ifdef CONFIG_PROC_FS
767int ip_misc_proc_init(void);
768#endif
769
David Brazdil0f672f62019-12-10 10:32:29 +0000770int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000771 struct netlink_ext_ack *extack);
772
Olivier Deprez0e641232021-09-23 10:07:05 +0200773static inline bool inetdev_valid_mtu(unsigned int mtu)
774{
775 return likely(mtu >= IPV4_MIN_MTU);
776}
777
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000778#endif /* _IP_H */