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
ipv4_l3mdev_skb(u16 flags)67 static inline bool ipv4_l3mdev_skb(u16 flags)
68 {
69 return !!(flags & IPSKB_L3SLAVE);
70 }
71
ip_hdrlen(const struct sk_buff * skb)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
ipcm_init(struct ipcm_cookie * ipcm)88 static inline void ipcm_init(struct ipcm_cookie *ipcm)
89 {
90 *ipcm = (struct ipcm_cookie) { .tos = -1 };
91 }
92
ipcm_init_sk(struct ipcm_cookie * ipcm,const struct inet_sock * inet)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 = READ_ONCE(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 */
inet_sdif(const struct sk_buff * skb)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
ip_fraglist_next(struct ip_fraglist_iter * iter)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
ip_finish_skb(struct sock * sk,struct flowi4 * fl4)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. */
ip_sendmsg_scope(const struct inet_sock * inet,const struct ipcm_cookie * ipc,const struct msghdr * msg)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_unsized *uaddr, int addr_len);
265 int ip4_datagram_connect(struct sock *sk, struct sockaddr_unsized *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
ip_reply_arg_flowi_flags(const struct ip_reply_arg * arg)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
snmp_get_cpu_field(void __percpu * mib,int cpu,int offt)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
snmp_get_cpu_field64(void __percpu * mib,int cpu,int offct,size_t syncp_offset)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
snmp_fold_field64(void __percpu * mib,int offt,size_t syncp_off)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_cnt(buff64, stats_list, cnt, \
330 mib_statistic, offset) \
331 { \
332 int i, c; \
333 for_each_possible_cpu(c) { \
334 for (i = 0; i < cnt; i++) \
335 buff64[i] += snmp_get_cpu_field64( \
336 mib_statistic, \
337 c, stats_list[i].entry, \
338 offset); \
339 } \
340 }
341
342 #define snmp_get_cpu_field_batch_cnt(buff, stats_list, cnt, mib_statistic) \
343 { \
344 int i, c; \
345 for_each_possible_cpu(c) { \
346 for (i = 0; i < cnt; i++) \
347 buff[i] += snmp_get_cpu_field( \
348 mib_statistic, \
349 c, stats_list[i].entry); \
350 } \
351 }
352
inet_get_local_port_range(const struct net * net,int * low,int * high)353 static inline void inet_get_local_port_range(const struct net *net, int *low, int *high)
354 {
355 u32 range = READ_ONCE(net->ipv4.ip_local_ports.range);
356
357 *low = range & 0xffff;
358 *high = range >> 16;
359 }
360 bool inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high);
361
362 #ifdef CONFIG_SYSCTL
inet_is_local_reserved_port(const struct net * net,unsigned short port)363 static inline bool inet_is_local_reserved_port(const struct net *net, unsigned short port)
364 {
365 if (!net->ipv4.sysctl_local_reserved_ports)
366 return false;
367 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
368 }
369
sysctl_dev_name_is_allowed(const char * name)370 static inline bool sysctl_dev_name_is_allowed(const char *name)
371 {
372 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
373 }
374
inet_port_requires_bind_service(struct net * net,unsigned short port)375 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
376 {
377 return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
378 }
379
380 #else
inet_is_local_reserved_port(struct net * net,unsigned short port)381 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
382 {
383 return false;
384 }
385
inet_port_requires_bind_service(struct net * net,unsigned short port)386 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
387 {
388 return port < PROT_SOCK;
389 }
390 #endif
391
392 __be32 inet_current_timestamp(void);
393
394 /* From inetpeer.c */
395 extern int inet_peer_threshold;
396 extern int inet_peer_minttl;
397 extern int inet_peer_maxttl;
398
399 void ipfrag_init(void);
400
401 void ip_static_sysctl_init(void);
402
403 #define IP4_REPLY_MARK(net, mark) \
404 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
405
ip_is_fragment(const struct iphdr * iph)406 static inline bool ip_is_fragment(const struct iphdr *iph)
407 {
408 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
409 }
410
411 #ifdef CONFIG_INET
412 #include <net/dst.h>
413
414 /* The function in 2.2 was invalid, producing wrong result for
415 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
416 static inline
ip_decrease_ttl(struct iphdr * iph)417 int ip_decrease_ttl(struct iphdr *iph)
418 {
419 u32 check = (__force u32)iph->check;
420 check += (__force u32)htons(0x0100);
421 iph->check = (__force __sum16)(check + (check>=0xFFFF));
422 return --iph->ttl;
423 }
424
ip4h_dscp(const struct iphdr * ip4h)425 static inline dscp_t ip4h_dscp(const struct iphdr *ip4h)
426 {
427 return inet_dsfield_to_dscp(ip4h->tos);
428 }
429
ip_mtu_locked(const struct dst_entry * dst)430 static inline int ip_mtu_locked(const struct dst_entry *dst)
431 {
432 const struct rtable *rt = dst_rtable(dst);
433
434 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
435 }
436
437 static inline
ip_dont_fragment(const struct sock * sk,const struct dst_entry * dst)438 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
439 {
440 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
441
442 return pmtudisc == IP_PMTUDISC_DO ||
443 (pmtudisc == IP_PMTUDISC_WANT &&
444 !ip_mtu_locked(dst));
445 }
446
ip_sk_accept_pmtu(const struct sock * sk)447 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
448 {
449 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
450
451 return pmtudisc != IP_PMTUDISC_INTERFACE &&
452 pmtudisc != IP_PMTUDISC_OMIT;
453 }
454
ip_sk_use_pmtu(const struct sock * sk)455 static inline bool ip_sk_use_pmtu(const struct sock *sk)
456 {
457 return READ_ONCE(inet_sk(sk)->pmtudisc) < IP_PMTUDISC_PROBE;
458 }
459
ip_sk_ignore_df(const struct sock * sk)460 static inline bool ip_sk_ignore_df(const struct sock *sk)
461 {
462 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
463
464 return pmtudisc < IP_PMTUDISC_DO || pmtudisc == IP_PMTUDISC_OMIT;
465 }
466
ip_dst_mtu_maybe_forward(const struct dst_entry * dst,bool forwarding)467 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
468 bool forwarding)
469 {
470 const struct rtable *rt = dst_rtable(dst);
471 const struct net_device *dev;
472 unsigned int mtu, res;
473 struct net *net;
474
475 rcu_read_lock();
476
477 dev = dst_dev_rcu(dst);
478 net = dev_net_rcu(dev);
479 if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
480 ip_mtu_locked(dst) ||
481 !forwarding) {
482 mtu = rt->rt_pmtu;
483 if (mtu && time_before(jiffies, READ_ONCE(rt->dst.expires)))
484 goto out;
485 }
486
487 /* 'forwarding = true' case should always honour route mtu */
488 mtu = dst_metric_raw(dst, RTAX_MTU);
489 if (mtu)
490 goto out;
491
492 mtu = READ_ONCE(dev->mtu);
493
494 if (unlikely(ip_mtu_locked(dst))) {
495 if (rt->rt_uses_gateway && mtu > 576)
496 mtu = 576;
497 }
498
499 out:
500 mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
501
502 res = mtu - lwtunnel_headroom(dst->lwtstate, mtu);
503
504 rcu_read_unlock();
505
506 return res;
507 }
508
509 /* Configured/administrative MTU of a route, for advertising the TCP MSS.
510 *
511 * Unlike ip_dst_mtu_maybe_forward(), this deliberately ignores the
512 * ICMP-learned path MTU (rt->rt_pmtu). The advertised MSS bounds what the
513 * peer may send to us and must reflect our receive capability (the device or
514 * route-configured MTU), not a path MTU learned on the reverse (send)
515 * direction, which may not apply to the peer->us path and outlives the fnhe
516 * for the whole connection. See RFC 2923 section 2.3 and the comment above
517 * tcp_advertise_mss().
518 */
ip_dst_mtu_configured(const struct dst_entry * dst)519 static inline unsigned int ip_dst_mtu_configured(const struct dst_entry *dst)
520 {
521 unsigned int mtu, res;
522
523 rcu_read_lock();
524 mtu = dst_metric_raw(dst, RTAX_MTU);
525 if (!mtu)
526 mtu = READ_ONCE(dst_dev_rcu(dst)->mtu);
527 mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
528 res = mtu - lwtunnel_headroom(dst->lwtstate, mtu);
529 rcu_read_unlock();
530
531 return res;
532 }
533
ip_skb_dst_mtu(struct sock * sk,const struct sk_buff * skb)534 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
535 const struct sk_buff *skb)
536 {
537 const struct dst_entry *dst = skb_dst(skb);
538 unsigned int mtu;
539
540 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
541 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
542
543 return ip_dst_mtu_maybe_forward(dst, forwarding);
544 }
545
546 mtu = min(READ_ONCE(dst_dev(dst)->mtu), IP_MAX_MTU);
547 return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
548 }
549
550 struct dst_metrics *ip_fib_metrics_init(struct nlattr *fc_mx, int fc_mx_len,
551 struct netlink_ext_ack *extack);
ip_fib_metrics_put(struct dst_metrics * fib_metrics)552 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
553 {
554 if (fib_metrics != &dst_default_metrics &&
555 refcount_dec_and_test(&fib_metrics->refcnt))
556 kfree(fib_metrics);
557 }
558
559 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
560 static inline
ip_dst_init_metrics(struct dst_entry * dst,struct dst_metrics * fib_metrics)561 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
562 {
563 dst_init_metrics(dst, fib_metrics->metrics, true);
564
565 if (fib_metrics != &dst_default_metrics) {
566 dst->_metrics |= DST_METRICS_REFCOUNTED;
567 refcount_inc(&fib_metrics->refcnt);
568 }
569 }
570
571 static inline
ip_dst_metrics_put(struct dst_entry * dst)572 void ip_dst_metrics_put(struct dst_entry *dst)
573 {
574 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
575
576 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
577 kfree(p);
578 }
579
580 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
581
ip_select_ident_segs(struct net * net,struct sk_buff * skb,struct sock * sk,int segs)582 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
583 struct sock *sk, int segs)
584 {
585 struct iphdr *iph = ip_hdr(skb);
586
587 /* We had many attacks based on IPID, use the private
588 * generator as much as we can.
589 */
590 if (sk && inet_sk(sk)->inet_daddr) {
591 int val;
592
593 /* avoid atomic operations for TCP,
594 * as we hold socket lock at this point.
595 */
596 if (sk_is_tcp(sk)) {
597 sock_owned_by_me(sk);
598 val = atomic_read(&inet_sk(sk)->inet_id);
599 atomic_set(&inet_sk(sk)->inet_id, val + segs);
600 } else {
601 val = atomic_add_return(segs, &inet_sk(sk)->inet_id);
602 }
603 iph->id = htons(val);
604 return;
605 }
606 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
607 iph->id = 0;
608 } else {
609 /* Unfortunately we need the big hammer to get a suitable IPID */
610 __ip_select_ident(net, iph, segs);
611 }
612 }
613
ip_select_ident(struct net * net,struct sk_buff * skb,struct sock * sk)614 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
615 struct sock *sk)
616 {
617 ip_select_ident_segs(net, skb, sk, 1);
618 }
619
inet_compute_pseudo(struct sk_buff * skb,int proto)620 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
621 {
622 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
623 skb->len, proto, 0);
624 }
625
626 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
627 * Equivalent to : flow->v4addrs.src = iph->saddr;
628 * flow->v4addrs.dst = iph->daddr;
629 */
iph_to_flow_copy_v4addrs(struct flow_keys * flow,const struct iphdr * iph)630 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
631 const struct iphdr *iph)
632 {
633 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
634 offsetof(typeof(flow->addrs), v4addrs.src) +
635 sizeof(flow->addrs.v4addrs.src));
636 memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs));
637 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
638 }
639
640 /*
641 * Map a multicast IP onto multicast MAC for type ethernet.
642 */
643
ip_eth_mc_map(__be32 naddr,char * buf)644 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
645 {
646 __u32 addr=ntohl(naddr);
647 buf[0]=0x01;
648 buf[1]=0x00;
649 buf[2]=0x5e;
650 buf[5]=addr&0xFF;
651 addr>>=8;
652 buf[4]=addr&0xFF;
653 addr>>=8;
654 buf[3]=addr&0x7F;
655 }
656
657 /*
658 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
659 * Leave P_Key as 0 to be filled in by driver.
660 */
661
ip_ib_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)662 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
663 {
664 __u32 addr;
665 unsigned char scope = broadcast[5] & 0xF;
666
667 buf[0] = 0; /* Reserved */
668 buf[1] = 0xff; /* Multicast QPN */
669 buf[2] = 0xff;
670 buf[3] = 0xff;
671 addr = ntohl(naddr);
672 buf[4] = 0xff;
673 buf[5] = 0x10 | scope; /* scope from broadcast address */
674 buf[6] = 0x40; /* IPv4 signature */
675 buf[7] = 0x1b;
676 buf[8] = broadcast[8]; /* P_Key */
677 buf[9] = broadcast[9];
678 buf[10] = 0;
679 buf[11] = 0;
680 buf[12] = 0;
681 buf[13] = 0;
682 buf[14] = 0;
683 buf[15] = 0;
684 buf[19] = addr & 0xff;
685 addr >>= 8;
686 buf[18] = addr & 0xff;
687 addr >>= 8;
688 buf[17] = addr & 0xff;
689 addr >>= 8;
690 buf[16] = addr & 0x0f;
691 }
692
ip_ipgre_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)693 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
694 {
695 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
696 memcpy(buf, broadcast, 4);
697 else
698 memcpy(buf, &naddr, sizeof(naddr));
699 }
700
701 #if IS_ENABLED(CONFIG_IPV6)
702 #include <linux/ipv6.h>
703 #endif
704
inet_reset_saddr(struct sock * sk)705 static __inline__ void inet_reset_saddr(struct sock *sk)
706 {
707 inet_sk(sk)->inet_saddr = 0;
708 WRITE_ONCE(inet_sk(sk)->inet_rcv_saddr, 0);
709 #if IS_ENABLED(CONFIG_IPV6)
710 if (sk->sk_family == PF_INET6) {
711 struct ipv6_pinfo *np = inet6_sk(sk);
712
713 memset(&np->saddr, 0, sizeof(np->saddr));
714 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
715 }
716 #endif
717 }
718
719 #endif
720
ipv4_addr_hash(__be32 ip)721 static inline unsigned int ipv4_addr_hash(__be32 ip)
722 {
723 return (__force unsigned int) ip;
724 }
725
__ipv4_addr_hash(const __be32 ip,const u32 initval)726 static inline u32 __ipv4_addr_hash(const __be32 ip, const u32 initval)
727 {
728 return jhash_1word((__force u32)ip, initval);
729 }
730
ipv4_portaddr_hash(const struct net * net,__be32 saddr,unsigned int port)731 static inline u32 ipv4_portaddr_hash(const struct net *net,
732 __be32 saddr,
733 unsigned int port)
734 {
735 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
736 }
737
738 bool ip_call_ra_chain(struct sk_buff *skb);
739
740 /*
741 * Functions provided by ip_fragment.c
742 */
743
744 enum ip_defrag_users {
745 IP_DEFRAG_LOCAL_DELIVER,
746 IP_DEFRAG_CALL_RA_CHAIN,
747 IP_DEFRAG_CONNTRACK_IN,
748 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
749 IP_DEFRAG_CONNTRACK_OUT,
750 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
751 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
752 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
753 IP_DEFRAG_VS_IN,
754 IP_DEFRAG_VS_OUT,
755 IP_DEFRAG_VS_FWD,
756 IP_DEFRAG_AF_PACKET,
757 IP_DEFRAG_MACVLAN,
758 };
759
760 /* Return true if the value of 'user' is between 'lower_bond'
761 * and 'upper_bond' inclusively.
762 */
ip_defrag_user_in_between(u32 user,enum ip_defrag_users lower_bond,enum ip_defrag_users upper_bond)763 static inline bool ip_defrag_user_in_between(u32 user,
764 enum ip_defrag_users lower_bond,
765 enum ip_defrag_users upper_bond)
766 {
767 return user >= lower_bond && user <= upper_bond;
768 }
769
770 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
771 #ifdef CONFIG_INET
772 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
773 #else
ip_check_defrag(struct net * net,struct sk_buff * skb,u32 user)774 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
775 {
776 return skb;
777 }
778 #endif
779
780 /*
781 * Functions provided by ip_forward.c
782 */
783
784 int ip_forward(struct sk_buff *skb);
785
786 /*
787 * Functions provided by ip_options.c
788 */
789
790 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
791 __be32 daddr, struct rtable *rt);
792
793 int __ip_options_echo(struct net *net, struct ip_options *dopt,
794 struct sk_buff *skb, const struct ip_options *sopt);
ip_options_echo(struct net * net,struct ip_options * dopt,struct sk_buff * skb)795 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
796 struct sk_buff *skb)
797 {
798 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
799 }
800
801 void ip_options_fragment(struct sk_buff *skb);
802 int __ip_options_compile(struct net *net, struct ip_options *opt,
803 struct sk_buff *skb, __be32 *info);
804 int ip_options_compile(struct net *net, struct ip_options *opt,
805 struct sk_buff *skb);
806 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
807 sockptr_t data, int optlen);
808 void ip_options_undo(struct ip_options *opt);
809 void ip_forward_options(struct sk_buff *skb);
810 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
811
812 /*
813 * Functions provided by ip_sockglue.c
814 */
815
816 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb, bool drop_dst);
817 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
818 struct sk_buff *skb, int tlen, int offset);
819 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
820 struct ipcm_cookie *ipc, bool allow_ipv6);
821 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl);
822 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
823 unsigned int optlen);
824 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
825 unsigned int optlen);
826 int do_ip_getsockopt(struct sock *sk, int level, int optname,
827 sockptr_t optval, sockptr_t optlen);
828 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
829 int __user *optlen);
830 int ip_ra_control(struct sock *sk, unsigned char on,
831 void (*destructor)(struct sock *));
832
833 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len);
834 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
835 u32 info, u8 *payload);
836 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
837 u32 info);
838
ip_cmsg_recv(struct msghdr * msg,struct sk_buff * skb)839 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
840 {
841 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
842 }
843
844 bool icmp_global_allow(struct net *net);
845 void icmp_global_consume(struct net *net);
846
847 #ifdef CONFIG_PROC_FS
848 int ip_misc_proc_init(void);
849 #endif
850
851 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
852 struct netlink_ext_ack *extack);
853
inetdev_valid_mtu(unsigned int mtu)854 static inline bool inetdev_valid_mtu(unsigned int mtu)
855 {
856 return likely(mtu >= IPV4_MIN_MTU);
857 }
858
859 void ip_sock_set_freebind(struct sock *sk);
860 int ip_sock_set_mtu_discover(struct sock *sk, int val);
861 void ip_sock_set_pktinfo(struct sock *sk);
862 void ip_sock_set_recverr(struct sock *sk);
863 void ip_sock_set_tos(struct sock *sk, int val);
864 void __ip_sock_set_tos(struct sock *sk, int val);
865
866 #endif /* _IP_H */
867