blob: de2dc22a78f938470ca5cf4d973adaa538a39137 [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>
Olivier Deprez157378f2022-04-04 15:47:50 +020026#include <linux/sockptr.h>
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000027
28#include <net/inet_sock.h>
29#include <net/route.h>
30#include <net/snmp.h>
31#include <net/flow.h>
32#include <net/flow_dissector.h>
33#include <net/netns/hash.h>
Olivier Deprez0e641232021-09-23 10:07:05 +020034#include <net/lwtunnel.h>
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000035
36#define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */
37#define IPV4_MIN_MTU 68 /* RFC 791 */
38
David Brazdil0f672f62019-12-10 10:32:29 +000039extern unsigned int sysctl_fib_sync_mem;
40extern unsigned int sysctl_fib_sync_mem_min;
41extern unsigned int sysctl_fib_sync_mem_max;
42
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000043struct sock;
44
45struct inet_skb_parm {
46 int iif;
47 struct ip_options opt; /* Compiled IP options */
48 u16 flags;
49
50#define IPSKB_FORWARDED BIT(0)
51#define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
52#define IPSKB_XFRM_TRANSFORMED BIT(2)
53#define IPSKB_FRAG_COMPLETE BIT(3)
54#define IPSKB_REROUTED BIT(4)
55#define IPSKB_DOREDIRECT BIT(5)
56#define IPSKB_FRAG_PMTU BIT(6)
57#define IPSKB_L3SLAVE BIT(7)
58
59 u16 frag_max_size;
60};
61
62static inline bool ipv4_l3mdev_skb(u16 flags)
63{
64 return !!(flags & IPSKB_L3SLAVE);
65}
66
67static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
68{
69 return ip_hdr(skb)->ihl * 4;
70}
71
72struct ipcm_cookie {
73 struct sockcm_cookie sockc;
74 __be32 addr;
75 int oif;
76 struct ip_options_rcu *opt;
77 __u8 ttl;
78 __s16 tos;
79 char priority;
80 __u16 gso_size;
81};
82
83static inline void ipcm_init(struct ipcm_cookie *ipcm)
84{
85 *ipcm = (struct ipcm_cookie) { .tos = -1 };
86}
87
88static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
89 const struct inet_sock *inet)
90{
91 ipcm_init(ipcm);
92
David Brazdil0f672f62019-12-10 10:32:29 +000093 ipcm->sockc.mark = inet->sk.sk_mark;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +000094 ipcm->sockc.tsflags = inet->sk.sk_tsflags;
95 ipcm->oif = inet->sk.sk_bound_dev_if;
96 ipcm->addr = inet->inet_saddr;
97}
98
99#define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
100#define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
101
102/* return enslaved device index if relevant */
103static inline int inet_sdif(struct sk_buff *skb)
104{
105#if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
106 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
107 return IPCB(skb)->iif;
108#endif
109 return 0;
110}
111
112/* Special input handler for packets caught by router alert option.
113 They are selected only by protocol field, and then processed likely
114 local ones; but only if someone wants them! Otherwise, router
115 not running rsvpd will kill RSVP.
116
117 It is user level problem, what it will make with them.
118 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
119 but receiver should be enough clever f.e. to forward mtrace requests,
120 sent to multicast group to reach destination designated router.
121 */
122
123struct ip_ra_chain {
124 struct ip_ra_chain __rcu *next;
125 struct sock *sk;
126 union {
127 void (*destructor)(struct sock *);
128 struct sock *saved_sk;
129 };
130 struct rcu_head rcu;
131};
132
133/* IP flags. */
134#define IP_CE 0x8000 /* Flag: "Congestion" */
135#define IP_DF 0x4000 /* Flag: "Don't Fragment" */
136#define IP_MF 0x2000 /* Flag: "More Fragments" */
137#define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
138
139#define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
140
141struct msghdr;
142struct net_device;
143struct packet_type;
144struct rtable;
145struct sockaddr;
146
147int igmp_mc_init(void);
148
149/*
150 * Functions provided by ip.c
151 */
152
153int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
154 __be32 saddr, __be32 daddr,
Olivier Deprez157378f2022-04-04 15:47:50 +0200155 struct ip_options_rcu *opt, u8 tos);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000156int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
157 struct net_device *orig_dev);
158void ip_list_rcv(struct list_head *head, struct packet_type *pt,
159 struct net_device *orig_dev);
160int ip_local_deliver(struct sk_buff *skb);
David Brazdil0f672f62019-12-10 10:32:29 +0000161void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000162int ip_mr_input(struct sk_buff *skb);
163int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
164int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
165int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
166 int (*output)(struct net *, struct sock *, struct sk_buff *));
David Brazdil0f672f62019-12-10 10:32:29 +0000167
168struct ip_fraglist_iter {
169 struct sk_buff *frag;
170 struct iphdr *iph;
171 int offset;
172 unsigned int hlen;
173};
174
175void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
176 unsigned int hlen, struct ip_fraglist_iter *iter);
177void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
178
179static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
180{
181 struct sk_buff *skb = iter->frag;
182
183 iter->frag = skb->next;
184 skb_mark_not_on_list(skb);
185
186 return skb;
187}
188
189struct ip_frag_state {
190 bool DF;
191 unsigned int hlen;
192 unsigned int ll_rs;
193 unsigned int mtu;
194 unsigned int left;
195 int offset;
196 int ptr;
197 __be16 not_last_frag;
198};
199
200void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
201 unsigned int mtu, bool DF, struct ip_frag_state *state);
202struct sk_buff *ip_frag_next(struct sk_buff *skb,
203 struct ip_frag_state *state);
204
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000205void ip_send_check(struct iphdr *ip);
206int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
207int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
208
209int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
210 __u8 tos);
211void ip_init(void);
212int ip_append_data(struct sock *sk, struct flowi4 *fl4,
213 int getfrag(void *from, char *to, int offset, int len,
214 int odd, struct sk_buff *skb),
215 void *from, int len, int protolen,
216 struct ipcm_cookie *ipc,
217 struct rtable **rt,
218 unsigned int flags);
219int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
220 struct sk_buff *skb);
221ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
222 int offset, size_t size, int flags);
223struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
224 struct sk_buff_head *queue,
225 struct inet_cork *cork);
226int ip_send_skb(struct net *net, struct sk_buff *skb);
227int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
228void ip_flush_pending_frames(struct sock *sk);
229struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
230 int getfrag(void *from, char *to, int offset,
231 int len, int odd, struct sk_buff *skb),
232 void *from, int length, int transhdrlen,
233 struct ipcm_cookie *ipc, struct rtable **rtp,
234 struct inet_cork *cork, unsigned int flags);
235
Olivier Deprez157378f2022-04-04 15:47:50 +0200236int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000237
238static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
239{
240 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
241}
242
243static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
244{
245 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
246}
247
248static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
249{
250 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
251}
252
253/* datagram.c */
254int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
255int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
256
257void ip4_datagram_release_cb(struct sock *sk);
258
259struct ip_reply_arg {
260 struct kvec iov[1];
261 int flags;
262 __wsum csum;
263 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
264 /* -1 if not needed */
265 int bound_dev_if;
266 u8 tos;
267 kuid_t uid;
268};
269
270#define IP_REPLY_ARG_NOSRCCHECK 1
271
272static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
273{
274 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
275}
276
277void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
278 const struct ip_options *sopt,
279 __be32 daddr, __be32 saddr,
280 const struct ip_reply_arg *arg,
David Brazdil0f672f62019-12-10 10:32:29 +0000281 unsigned int len, u64 transmit_time);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000282
283#define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
284#define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
285#define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
286#define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
287#define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
288#define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
289#define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
290#define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
291#define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
292#define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
293
294u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
295unsigned long snmp_fold_field(void __percpu *mib, int offt);
296#if BITS_PER_LONG==32
297u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
298 size_t syncp_offset);
299u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
300#else
301static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
302 size_t syncp_offset)
303{
304 return snmp_get_cpu_field(mib, cpu, offct);
305
306}
307
308static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
309{
310 return snmp_fold_field(mib, offt);
311}
312#endif
313
314#define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
315{ \
316 int i, c; \
317 for_each_possible_cpu(c) { \
318 for (i = 0; stats_list[i].name; i++) \
319 buff64[i] += snmp_get_cpu_field64( \
320 mib_statistic, \
321 c, stats_list[i].entry, \
322 offset); \
323 } \
324}
325
326#define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
327{ \
328 int i, c; \
329 for_each_possible_cpu(c) { \
330 for (i = 0; stats_list[i].name; i++) \
331 buff[i] += snmp_get_cpu_field( \
332 mib_statistic, \
333 c, stats_list[i].entry); \
334 } \
335}
336
337void inet_get_local_port_range(struct net *net, int *low, int *high);
338
339#ifdef CONFIG_SYSCTL
Olivier Deprez157378f2022-04-04 15:47:50 +0200340static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000341{
342 if (!net->ipv4.sysctl_local_reserved_ports)
Olivier Deprez157378f2022-04-04 15:47:50 +0200343 return false;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000344 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
345}
346
347static inline bool sysctl_dev_name_is_allowed(const char *name)
348{
349 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
350}
351
Olivier Deprez157378f2022-04-04 15:47:50 +0200352static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000353{
Olivier Deprez157378f2022-04-04 15:47:50 +0200354 return port < net->ipv4.sysctl_ip_prot_sock;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000355}
356
357#else
Olivier Deprez157378f2022-04-04 15:47:50 +0200358static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000359{
Olivier Deprez157378f2022-04-04 15:47:50 +0200360 return false;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000361}
362
Olivier Deprez157378f2022-04-04 15:47:50 +0200363static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000364{
Olivier Deprez157378f2022-04-04 15:47:50 +0200365 return port < PROT_SOCK;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000366}
367#endif
368
369__be32 inet_current_timestamp(void);
370
371/* From inetpeer.c */
372extern int inet_peer_threshold;
373extern int inet_peer_minttl;
374extern int inet_peer_maxttl;
375
376void ipfrag_init(void);
377
378void ip_static_sysctl_init(void);
379
380#define IP4_REPLY_MARK(net, mark) \
381 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
382
383static inline bool ip_is_fragment(const struct iphdr *iph)
384{
385 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
386}
387
388#ifdef CONFIG_INET
389#include <net/dst.h>
390
391/* The function in 2.2 was invalid, producing wrong result for
392 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
393static inline
394int ip_decrease_ttl(struct iphdr *iph)
395{
396 u32 check = (__force u32)iph->check;
397 check += (__force u32)htons(0x0100);
398 iph->check = (__force __sum16)(check + (check>=0xFFFF));
399 return --iph->ttl;
400}
401
402static inline int ip_mtu_locked(const struct dst_entry *dst)
403{
404 const struct rtable *rt = (const struct rtable *)dst;
405
406 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
407}
408
409static inline
410int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
411{
412 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
413
414 return pmtudisc == IP_PMTUDISC_DO ||
415 (pmtudisc == IP_PMTUDISC_WANT &&
416 !ip_mtu_locked(dst));
417}
418
419static inline bool ip_sk_accept_pmtu(const struct sock *sk)
420{
421 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
422 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
423}
424
425static inline bool ip_sk_use_pmtu(const struct sock *sk)
426{
427 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
428}
429
430static inline bool ip_sk_ignore_df(const struct sock *sk)
431{
432 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
433 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
434}
435
436static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
437 bool forwarding)
438{
439 struct net *net = dev_net(dst->dev);
Olivier Deprez0e641232021-09-23 10:07:05 +0200440 unsigned int mtu;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000441
442 if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
443 ip_mtu_locked(dst) ||
444 !forwarding)
445 return dst_mtu(dst);
446
Olivier Deprez0e641232021-09-23 10:07:05 +0200447 /* 'forwarding = true' case should always honour route mtu */
448 mtu = dst_metric_raw(dst, RTAX_MTU);
449 if (!mtu)
450 mtu = min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU);
451
452 return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000453}
454
455static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
456 const struct sk_buff *skb)
457{
Olivier Deprez0e641232021-09-23 10:07:05 +0200458 unsigned int mtu;
459
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000460 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
461 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
462
463 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
464 }
465
Olivier Deprez0e641232021-09-23 10:07:05 +0200466 mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
467 return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000468}
469
David Brazdil0f672f62019-12-10 10:32:29 +0000470struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx,
471 int fc_mx_len,
472 struct netlink_ext_ack *extack);
473static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
474{
475 if (fib_metrics != &dst_default_metrics &&
476 refcount_dec_and_test(&fib_metrics->refcnt))
477 kfree(fib_metrics);
478}
479
480/* ipv4 and ipv6 both use refcounted metrics if it is not the default */
481static inline
482void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
483{
484 dst_init_metrics(dst, fib_metrics->metrics, true);
485
486 if (fib_metrics != &dst_default_metrics) {
487 dst->_metrics |= DST_METRICS_REFCOUNTED;
488 refcount_inc(&fib_metrics->refcnt);
489 }
490}
491
492static inline
493void ip_dst_metrics_put(struct dst_entry *dst)
494{
495 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
496
497 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
498 kfree(p);
499}
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000500
501u32 ip_idents_reserve(u32 hash, int segs);
502void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
503
504static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
505 struct sock *sk, int segs)
506{
507 struct iphdr *iph = ip_hdr(skb);
508
Olivier Deprez157378f2022-04-04 15:47:50 +0200509 /* We had many attacks based on IPID, use the private
510 * generator as much as we can.
511 */
512 if (sk && inet_sk(sk)->inet_daddr) {
513 iph->id = htons(inet_sk(sk)->inet_id);
514 inet_sk(sk)->inet_id += segs;
515 return;
516 }
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000517 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
Olivier Deprez157378f2022-04-04 15:47:50 +0200518 iph->id = 0;
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000519 } else {
Olivier Deprez157378f2022-04-04 15:47:50 +0200520 /* Unfortunately we need the big hammer to get a suitable IPID */
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000521 __ip_select_ident(net, iph, segs);
522 }
523}
524
525static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
526 struct sock *sk)
527{
528 ip_select_ident_segs(net, skb, sk, 1);
529}
530
531static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
532{
533 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
534 skb->len, proto, 0);
535}
536
537/* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
538 * Equivalent to : flow->v4addrs.src = iph->saddr;
539 * flow->v4addrs.dst = iph->daddr;
540 */
541static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
542 const struct iphdr *iph)
543{
544 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
545 offsetof(typeof(flow->addrs), v4addrs.src) +
546 sizeof(flow->addrs.v4addrs.src));
547 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs));
548 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
549}
550
551static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
552{
553 const struct iphdr *iph = skb_gro_network_header(skb);
554
555 return csum_tcpudp_nofold(iph->saddr, iph->daddr,
556 skb_gro_len(skb), proto, 0);
557}
558
559/*
560 * Map a multicast IP onto multicast MAC for type ethernet.
561 */
562
563static inline void ip_eth_mc_map(__be32 naddr, char *buf)
564{
565 __u32 addr=ntohl(naddr);
566 buf[0]=0x01;
567 buf[1]=0x00;
568 buf[2]=0x5e;
569 buf[5]=addr&0xFF;
570 addr>>=8;
571 buf[4]=addr&0xFF;
572 addr>>=8;
573 buf[3]=addr&0x7F;
574}
575
576/*
577 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
578 * Leave P_Key as 0 to be filled in by driver.
579 */
580
581static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
582{
583 __u32 addr;
584 unsigned char scope = broadcast[5] & 0xF;
585
586 buf[0] = 0; /* Reserved */
587 buf[1] = 0xff; /* Multicast QPN */
588 buf[2] = 0xff;
589 buf[3] = 0xff;
590 addr = ntohl(naddr);
591 buf[4] = 0xff;
592 buf[5] = 0x10 | scope; /* scope from broadcast address */
593 buf[6] = 0x40; /* IPv4 signature */
594 buf[7] = 0x1b;
595 buf[8] = broadcast[8]; /* P_Key */
596 buf[9] = broadcast[9];
597 buf[10] = 0;
598 buf[11] = 0;
599 buf[12] = 0;
600 buf[13] = 0;
601 buf[14] = 0;
602 buf[15] = 0;
603 buf[19] = addr & 0xff;
604 addr >>= 8;
605 buf[18] = addr & 0xff;
606 addr >>= 8;
607 buf[17] = addr & 0xff;
608 addr >>= 8;
609 buf[16] = addr & 0x0f;
610}
611
612static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
613{
614 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
615 memcpy(buf, broadcast, 4);
616 else
617 memcpy(buf, &naddr, sizeof(naddr));
618}
619
620#if IS_ENABLED(CONFIG_IPV6)
621#include <linux/ipv6.h>
622#endif
623
624static __inline__ void inet_reset_saddr(struct sock *sk)
625{
626 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
627#if IS_ENABLED(CONFIG_IPV6)
628 if (sk->sk_family == PF_INET6) {
629 struct ipv6_pinfo *np = inet6_sk(sk);
630
631 memset(&np->saddr, 0, sizeof(np->saddr));
632 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
633 }
634#endif
635}
636
637#endif
638
639static inline unsigned int ipv4_addr_hash(__be32 ip)
640{
641 return (__force unsigned int) ip;
642}
643
644static inline u32 ipv4_portaddr_hash(const struct net *net,
645 __be32 saddr,
646 unsigned int port)
647{
648 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
649}
650
651bool ip_call_ra_chain(struct sk_buff *skb);
652
653/*
654 * Functions provided by ip_fragment.c
655 */
656
657enum ip_defrag_users {
658 IP_DEFRAG_LOCAL_DELIVER,
659 IP_DEFRAG_CALL_RA_CHAIN,
660 IP_DEFRAG_CONNTRACK_IN,
661 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
662 IP_DEFRAG_CONNTRACK_OUT,
663 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
664 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
665 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
666 IP_DEFRAG_VS_IN,
667 IP_DEFRAG_VS_OUT,
668 IP_DEFRAG_VS_FWD,
669 IP_DEFRAG_AF_PACKET,
670 IP_DEFRAG_MACVLAN,
671};
672
673/* Return true if the value of 'user' is between 'lower_bond'
674 * and 'upper_bond' inclusively.
675 */
676static inline bool ip_defrag_user_in_between(u32 user,
677 enum ip_defrag_users lower_bond,
678 enum ip_defrag_users upper_bond)
679{
680 return user >= lower_bond && user <= upper_bond;
681}
682
683int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
684#ifdef CONFIG_INET
685struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
686#else
687static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
688{
689 return skb;
690}
691#endif
692
693/*
694 * Functions provided by ip_forward.c
695 */
696
697int ip_forward(struct sk_buff *skb);
698
699/*
700 * Functions provided by ip_options.c
701 */
702
703void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
704 __be32 daddr, struct rtable *rt, int is_frag);
705
706int __ip_options_echo(struct net *net, struct ip_options *dopt,
707 struct sk_buff *skb, const struct ip_options *sopt);
708static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
709 struct sk_buff *skb)
710{
711 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
712}
713
714void ip_options_fragment(struct sk_buff *skb);
David Brazdil0f672f62019-12-10 10:32:29 +0000715int __ip_options_compile(struct net *net, struct ip_options *opt,
716 struct sk_buff *skb, __be32 *info);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000717int ip_options_compile(struct net *net, struct ip_options *opt,
718 struct sk_buff *skb);
719int ip_options_get(struct net *net, struct ip_options_rcu **optp,
Olivier Deprez157378f2022-04-04 15:47:50 +0200720 sockptr_t data, int optlen);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000721void ip_options_undo(struct ip_options *opt);
722void ip_forward_options(struct sk_buff *skb);
David Brazdil0f672f62019-12-10 10:32:29 +0000723int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000724
725/*
726 * Functions provided by ip_sockglue.c
727 */
728
729void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
730void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
731 struct sk_buff *skb, int tlen, int offset);
732int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
733 struct ipcm_cookie *ipc, bool allow_ipv6);
Olivier Deprez157378f2022-04-04 15:47:50 +0200734int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000735 unsigned int optlen);
736int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
737 int __user *optlen);
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000738int ip_ra_control(struct sock *sk, unsigned char on,
739 void (*destructor)(struct sock *));
740
741int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
742void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
743 u32 info, u8 *payload);
744void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
745 u32 info);
746
747static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
748{
749 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
750}
751
752bool icmp_global_allow(void);
753extern int sysctl_icmp_msgs_per_sec;
754extern int sysctl_icmp_msgs_burst;
755
756#ifdef CONFIG_PROC_FS
757int ip_misc_proc_init(void);
758#endif
759
David Brazdil0f672f62019-12-10 10:32:29 +0000760int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000761 struct netlink_ext_ack *extack);
762
Olivier Deprez0e641232021-09-23 10:07:05 +0200763static inline bool inetdev_valid_mtu(unsigned int mtu)
764{
765 return likely(mtu >= IPV4_MIN_MTU);
766}
767
Olivier Deprez157378f2022-04-04 15:47:50 +0200768void ip_sock_set_freebind(struct sock *sk);
769int ip_sock_set_mtu_discover(struct sock *sk, int val);
770void ip_sock_set_pktinfo(struct sock *sk);
771void ip_sock_set_recverr(struct sock *sk);
772void ip_sock_set_tos(struct sock *sk, int val);
773
Andrew Scullb4b6d4a2019-01-02 15:54:55 +0000774#endif /* _IP_H */