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
ip_check_csum_start(const struct sk_buff * skb)77 static inline int ip_check_csum_start(const struct sk_buff *skb)
78 {
79 if (unlikely(skb->csum_start < skb->network_header + ip_hdrlen(skb)))
80 return -EINVAL;
81 return 0;
82 }
83
84 struct ipcm_cookie {
85 struct sockcm_cookie sockc;
86 __be32 addr;
87 int oif;
88 struct ip_options_rcu *opt;
89 __u8 protocol;
90 __u8 ttl;
91 __s16 tos;
92 __u16 gso_size;
93 };
94
ipcm_init(struct ipcm_cookie * ipcm)95 static inline void ipcm_init(struct ipcm_cookie *ipcm)
96 {
97 *ipcm = (struct ipcm_cookie) { .tos = -1 };
98 }
99
ipcm_init_sk(struct ipcm_cookie * ipcm,const struct inet_sock * inet)100 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
101 const struct inet_sock *inet)
102 {
103 *ipcm = (struct ipcm_cookie) {
104 .tos = READ_ONCE(inet->tos),
105 };
106
107 sockcm_init(&ipcm->sockc, &inet->sk);
108
109 ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if);
110 ipcm->addr = inet->inet_saddr;
111 ipcm->protocol = READ_ONCE(inet->inet_num);
112 }
113
114 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
115 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
116
117 /* return enslaved device index if relevant */
inet_sdif(const struct sk_buff * skb)118 static inline int inet_sdif(const struct sk_buff *skb)
119 {
120 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
121 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
122 return IPCB(skb)->iif;
123 #endif
124 return 0;
125 }
126
127 /* Special input handler for packets caught by router alert option.
128 They are selected only by protocol field, and then processed likely
129 local ones; but only if someone wants them! Otherwise, router
130 not running rsvpd will kill RSVP.
131
132 It is user level problem, what it will make with them.
133 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
134 but receiver should be enough clever f.e. to forward mtrace requests,
135 sent to multicast group to reach destination designated router.
136 */
137
138 struct ip_ra_chain {
139 struct ip_ra_chain __rcu *next;
140 struct sock *sk;
141 union {
142 void (*destructor)(struct sock *);
143 struct sock *saved_sk;
144 };
145 struct rcu_head rcu;
146 };
147
148 /* IP flags. */
149 #define IP_CE 0x8000 /* Flag: "Congestion" */
150 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
151 #define IP_MF 0x2000 /* Flag: "More Fragments" */
152 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
153
154 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
155
156 struct msghdr;
157 struct net_device;
158 struct packet_type;
159 struct rtable;
160 struct sockaddr;
161
162 int igmp_mc_init(void);
163
164 /*
165 * Functions provided by ip.c
166 */
167
168 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
169 __be32 saddr, __be32 daddr,
170 struct ip_options_rcu *opt, u8 tos);
171 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
172 struct net_device *orig_dev);
173 void ip_list_rcv(struct list_head *head, struct packet_type *pt,
174 struct net_device *orig_dev);
175 int ip_local_deliver(struct sk_buff *skb);
176 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
177 int ip_mr_input(struct sk_buff *skb);
178 int ip_mr_output(struct net *net, struct sock *sk, struct sk_buff *skb);
179 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
180 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
181 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
182 int (*output)(struct net *, struct sock *, struct sk_buff *));
183
184 struct ip_fraglist_iter {
185 struct sk_buff *frag;
186 struct iphdr *iph;
187 int offset;
188 unsigned int hlen;
189 };
190
191 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
192 unsigned int hlen, struct ip_fraglist_iter *iter);
193 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
194
ip_fraglist_next(struct ip_fraglist_iter * iter)195 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
196 {
197 struct sk_buff *skb = iter->frag;
198
199 iter->frag = skb->next;
200 skb_mark_not_on_list(skb);
201
202 return skb;
203 }
204
205 struct ip_frag_state {
206 bool DF;
207 unsigned int hlen;
208 unsigned int ll_rs;
209 unsigned int mtu;
210 unsigned int left;
211 int offset;
212 int ptr;
213 __be16 not_last_frag;
214 };
215
216 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
217 unsigned int mtu, bool DF, struct ip_frag_state *state);
218 struct sk_buff *ip_frag_next(struct sk_buff *skb,
219 struct ip_frag_state *state);
220
221 void ip_send_check(struct iphdr *ip);
222 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
223 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
224
225 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
226 __u8 tos);
227 void ip_init(void);
228 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
229 int getfrag(void *from, char *to, int offset, int len,
230 int odd, struct sk_buff *skb),
231 void *from, int len, int protolen,
232 struct ipcm_cookie *ipc,
233 struct rtable **rt,
234 unsigned int flags);
235 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
236 struct sk_buff *skb);
237 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
238 struct sk_buff_head *queue,
239 struct inet_cork *cork);
240 int ip_send_skb(struct net *net, struct sk_buff *skb);
241 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
242 void ip_flush_pending_frames(struct sock *sk);
243 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
244 int getfrag(void *from, char *to, int offset,
245 int len, int odd, struct sk_buff *skb),
246 void *from, int length, int transhdrlen,
247 struct ipcm_cookie *ipc, struct rtable **rtp,
248 struct inet_cork *cork, unsigned int flags);
249
250 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
251
ip_finish_skb(struct sock * sk,struct flowi4 * fl4)252 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
253 {
254 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
255 }
256
257 /* 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)258 static inline u8 ip_sendmsg_scope(const struct inet_sock *inet,
259 const struct ipcm_cookie *ipc,
260 const struct msghdr *msg)
261 {
262 if (sock_flag(&inet->sk, SOCK_LOCALROUTE) ||
263 msg->msg_flags & MSG_DONTROUTE ||
264 (ipc->opt && ipc->opt->opt.is_strictroute))
265 return RT_SCOPE_LINK;
266
267 return RT_SCOPE_UNIVERSE;
268 }
269
270 /* datagram.c */
271 int __ip4_datagram_connect(struct sock *sk, struct sockaddr_unsized *uaddr, int addr_len);
272 int ip4_datagram_connect(struct sock *sk, struct sockaddr_unsized *uaddr, int addr_len);
273
274 void ip4_datagram_release_cb(struct sock *sk);
275
276 struct ip_reply_arg {
277 struct kvec iov[1];
278 int flags;
279 __wsum csum;
280 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
281 /* -1 if not needed */
282 int bound_dev_if;
283 u8 tos;
284 kuid_t uid;
285 };
286
287 #define IP_REPLY_ARG_NOSRCCHECK 1
288
ip_reply_arg_flowi_flags(const struct ip_reply_arg * arg)289 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
290 {
291 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
292 }
293
294 void ip_send_unicast_reply(struct sock *sk, const struct sock *orig_sk,
295 struct sk_buff *skb,
296 const struct ip_options *sopt,
297 __be32 daddr, __be32 saddr,
298 const struct ip_reply_arg *arg,
299 unsigned int len, u64 transmit_time, u32 txhash);
300
301 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
302 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
303 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
304 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
305 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
306 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
307 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
308 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
309 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
310 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
311
snmp_get_cpu_field(void __percpu * mib,int cpu,int offt)312 static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt)
313 {
314 return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt);
315 }
316
317 unsigned long snmp_fold_field(void __percpu *mib, int offt);
318 #if BITS_PER_LONG==32
319 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
320 size_t syncp_offset);
321 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
322 #else
snmp_get_cpu_field64(void __percpu * mib,int cpu,int offct,size_t syncp_offset)323 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
324 size_t syncp_offset)
325 {
326 return snmp_get_cpu_field(mib, cpu, offct);
327
328 }
329
snmp_fold_field64(void __percpu * mib,int offt,size_t syncp_off)330 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
331 {
332 return snmp_fold_field(mib, offt);
333 }
334 #endif
335
336 #define snmp_get_cpu_field64_batch_cnt(buff64, stats_list, cnt, \
337 mib_statistic, offset) \
338 { \
339 int i, c; \
340 for_each_possible_cpu(c) { \
341 for (i = 0; i < cnt; i++) \
342 buff64[i] += snmp_get_cpu_field64( \
343 mib_statistic, \
344 c, stats_list[i].entry, \
345 offset); \
346 } \
347 }
348
349 #define snmp_get_cpu_field_batch_cnt(buff, stats_list, cnt, mib_statistic) \
350 { \
351 int i, c; \
352 for_each_possible_cpu(c) { \
353 for (i = 0; i < cnt; i++) \
354 buff[i] += snmp_get_cpu_field( \
355 mib_statistic, \
356 c, stats_list[i].entry); \
357 } \
358 }
359
inet_get_local_port_range(const struct net * net,int * low,int * high)360 static inline void inet_get_local_port_range(const struct net *net, int *low, int *high)
361 {
362 u32 range = READ_ONCE(net->ipv4.ip_local_ports.range);
363
364 *low = range & 0xffff;
365 *high = range >> 16;
366 }
367 bool inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high);
368
369 #ifdef CONFIG_SYSCTL
inet_is_local_reserved_port(const struct net * net,unsigned short port)370 static inline bool inet_is_local_reserved_port(const struct net *net, unsigned short port)
371 {
372 if (!net->ipv4.sysctl_local_reserved_ports)
373 return false;
374 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
375 }
376
sysctl_dev_name_is_allowed(const char * name)377 static inline bool sysctl_dev_name_is_allowed(const char *name)
378 {
379 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
380 }
381
inet_port_requires_bind_service(struct net * net,unsigned short port)382 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
383 {
384 return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
385 }
386
387 #else
inet_is_local_reserved_port(struct net * net,unsigned short port)388 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
389 {
390 return false;
391 }
392
inet_port_requires_bind_service(struct net * net,unsigned short port)393 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
394 {
395 return port < PROT_SOCK;
396 }
397 #endif
398
399 __be32 inet_current_timestamp(void);
400
401 /* From inetpeer.c */
402 extern int inet_peer_threshold;
403 extern int inet_peer_minttl;
404 extern int inet_peer_maxttl;
405
406 void ipfrag_init(void);
407
408 void ip_static_sysctl_init(void);
409
410 #define IP4_REPLY_MARK(net, mark) \
411 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
412
ip_is_fragment(const struct iphdr * iph)413 static inline bool ip_is_fragment(const struct iphdr *iph)
414 {
415 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
416 }
417
418 #ifdef CONFIG_INET
419 #include <net/dst.h>
420
421 /* The function in 2.2 was invalid, producing wrong result for
422 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
423 static inline
ip_decrease_ttl(struct iphdr * iph)424 int ip_decrease_ttl(struct iphdr *iph)
425 {
426 u32 check = (__force u32)iph->check;
427 check += (__force u32)htons(0x0100);
428 iph->check = (__force __sum16)(check + (check>=0xFFFF));
429 return --iph->ttl;
430 }
431
ip4h_dscp(const struct iphdr * ip4h)432 static inline dscp_t ip4h_dscp(const struct iphdr *ip4h)
433 {
434 return inet_dsfield_to_dscp(ip4h->tos);
435 }
436
ip_mtu_locked(const struct dst_entry * dst)437 static inline int ip_mtu_locked(const struct dst_entry *dst)
438 {
439 const struct rtable *rt = dst_rtable(dst);
440
441 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
442 }
443
444 static inline
ip_dont_fragment(const struct sock * sk,const struct dst_entry * dst)445 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
446 {
447 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
448
449 return pmtudisc == IP_PMTUDISC_DO ||
450 (pmtudisc == IP_PMTUDISC_WANT &&
451 !ip_mtu_locked(dst));
452 }
453
ip_sk_accept_pmtu(const struct sock * sk)454 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
455 {
456 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
457
458 return pmtudisc != IP_PMTUDISC_INTERFACE &&
459 pmtudisc != IP_PMTUDISC_OMIT;
460 }
461
ip_sk_use_pmtu(const struct sock * sk)462 static inline bool ip_sk_use_pmtu(const struct sock *sk)
463 {
464 return READ_ONCE(inet_sk(sk)->pmtudisc) < IP_PMTUDISC_PROBE;
465 }
466
ip_sk_ignore_df(const struct sock * sk)467 static inline bool ip_sk_ignore_df(const struct sock *sk)
468 {
469 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
470
471 return pmtudisc < IP_PMTUDISC_DO || pmtudisc == IP_PMTUDISC_OMIT;
472 }
473
ip_dst_mtu_maybe_forward(const struct dst_entry * dst,bool forwarding)474 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
475 bool forwarding)
476 {
477 const struct rtable *rt = dst_rtable(dst);
478 const struct net_device *dev;
479 unsigned int mtu, res;
480 struct net *net;
481
482 rcu_read_lock();
483
484 dev = dst_dev_rcu(dst);
485 net = dev_net_rcu(dev);
486 if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
487 ip_mtu_locked(dst) ||
488 !forwarding) {
489 mtu = rt->rt_pmtu;
490 if (mtu && time_before(jiffies, READ_ONCE(rt->dst.expires)))
491 goto out;
492 }
493
494 /* 'forwarding = true' case should always honour route mtu */
495 mtu = dst_metric_raw(dst, RTAX_MTU);
496 if (mtu)
497 goto out;
498
499 mtu = READ_ONCE(dev->mtu);
500
501 if (unlikely(ip_mtu_locked(dst))) {
502 if (rt->rt_uses_gateway && mtu > 576)
503 mtu = 576;
504 }
505
506 out:
507 mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
508
509 res = mtu - lwtunnel_headroom(dst->lwtstate, mtu);
510
511 rcu_read_unlock();
512
513 return res;
514 }
515
516 /* Configured/administrative MTU of a route, for advertising the TCP MSS.
517 *
518 * Unlike ip_dst_mtu_maybe_forward(), this deliberately ignores the
519 * ICMP-learned path MTU (rt->rt_pmtu). The advertised MSS bounds what the
520 * peer may send to us and must reflect our receive capability (the device or
521 * route-configured MTU), not a path MTU learned on the reverse (send)
522 * direction, which may not apply to the peer->us path and outlives the fnhe
523 * for the whole connection. See RFC 2923 section 2.3 and the comment above
524 * tcp_advertise_mss().
525 */
ip_dst_mtu_configured(const struct dst_entry * dst)526 static inline unsigned int ip_dst_mtu_configured(const struct dst_entry *dst)
527 {
528 unsigned int mtu, res;
529
530 rcu_read_lock();
531 mtu = dst_metric_raw(dst, RTAX_MTU);
532 if (!mtu)
533 mtu = READ_ONCE(dst_dev_rcu(dst)->mtu);
534 mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
535 res = mtu - lwtunnel_headroom(dst->lwtstate, mtu);
536 rcu_read_unlock();
537
538 return res;
539 }
540
ip_skb_dst_mtu(struct sock * sk,const struct sk_buff * skb)541 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
542 const struct sk_buff *skb)
543 {
544 const struct dst_entry *dst = skb_dst(skb);
545 unsigned int mtu;
546
547 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
548 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
549
550 return ip_dst_mtu_maybe_forward(dst, forwarding);
551 }
552
553 mtu = min(READ_ONCE(dst_dev(dst)->mtu), IP_MAX_MTU);
554 return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
555 }
556
557 struct dst_metrics *ip_fib_metrics_init(struct nlattr *fc_mx, int fc_mx_len,
558 struct netlink_ext_ack *extack);
ip_fib_metrics_put(struct dst_metrics * fib_metrics)559 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
560 {
561 if (fib_metrics != &dst_default_metrics &&
562 refcount_dec_and_test(&fib_metrics->refcnt))
563 kfree(fib_metrics);
564 }
565
566 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
567 static inline
ip_dst_init_metrics(struct dst_entry * dst,struct dst_metrics * fib_metrics)568 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
569 {
570 dst_init_metrics(dst, fib_metrics->metrics, true);
571
572 if (fib_metrics != &dst_default_metrics) {
573 dst->_metrics |= DST_METRICS_REFCOUNTED;
574 refcount_inc(&fib_metrics->refcnt);
575 }
576 }
577
578 static inline
ip_dst_metrics_put(struct dst_entry * dst)579 void ip_dst_metrics_put(struct dst_entry *dst)
580 {
581 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
582
583 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
584 kfree(p);
585 }
586
587 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
588
ip_select_ident_segs(struct net * net,struct sk_buff * skb,struct sock * sk,int segs)589 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
590 struct sock *sk, int segs)
591 {
592 struct iphdr *iph = ip_hdr(skb);
593
594 /* We had many attacks based on IPID, use the private
595 * generator as much as we can.
596 */
597 if (sk && inet_sk(sk)->inet_daddr) {
598 int val;
599
600 /* avoid atomic operations for TCP,
601 * as we hold socket lock at this point.
602 */
603 if (sk_is_tcp(sk)) {
604 sock_owned_by_me(sk);
605 val = atomic_read(&inet_sk(sk)->inet_id);
606 atomic_set(&inet_sk(sk)->inet_id, val + segs);
607 } else {
608 val = atomic_add_return(segs, &inet_sk(sk)->inet_id);
609 }
610 iph->id = htons(val);
611 return;
612 }
613 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
614 iph->id = 0;
615 } else {
616 /* Unfortunately we need the big hammer to get a suitable IPID */
617 __ip_select_ident(net, iph, segs);
618 }
619 }
620
ip_select_ident(struct net * net,struct sk_buff * skb,struct sock * sk)621 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
622 struct sock *sk)
623 {
624 ip_select_ident_segs(net, skb, sk, 1);
625 }
626
inet_compute_pseudo(struct sk_buff * skb,int proto)627 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
628 {
629 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
630 skb->len, proto, 0);
631 }
632
633 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
634 * Equivalent to : flow->v4addrs.src = iph->saddr;
635 * flow->v4addrs.dst = iph->daddr;
636 */
iph_to_flow_copy_v4addrs(struct flow_keys * flow,const struct iphdr * iph)637 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
638 const struct iphdr *iph)
639 {
640 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
641 offsetof(typeof(flow->addrs), v4addrs.src) +
642 sizeof(flow->addrs.v4addrs.src));
643 memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs));
644 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
645 }
646
647 /*
648 * Map a multicast IP onto multicast MAC for type ethernet.
649 */
650
ip_eth_mc_map(__be32 naddr,char * buf)651 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
652 {
653 __u32 addr=ntohl(naddr);
654 buf[0]=0x01;
655 buf[1]=0x00;
656 buf[2]=0x5e;
657 buf[5]=addr&0xFF;
658 addr>>=8;
659 buf[4]=addr&0xFF;
660 addr>>=8;
661 buf[3]=addr&0x7F;
662 }
663
664 /*
665 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
666 * Leave P_Key as 0 to be filled in by driver.
667 */
668
ip_ib_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)669 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
670 {
671 __u32 addr;
672 unsigned char scope = broadcast[5] & 0xF;
673
674 buf[0] = 0; /* Reserved */
675 buf[1] = 0xff; /* Multicast QPN */
676 buf[2] = 0xff;
677 buf[3] = 0xff;
678 addr = ntohl(naddr);
679 buf[4] = 0xff;
680 buf[5] = 0x10 | scope; /* scope from broadcast address */
681 buf[6] = 0x40; /* IPv4 signature */
682 buf[7] = 0x1b;
683 buf[8] = broadcast[8]; /* P_Key */
684 buf[9] = broadcast[9];
685 buf[10] = 0;
686 buf[11] = 0;
687 buf[12] = 0;
688 buf[13] = 0;
689 buf[14] = 0;
690 buf[15] = 0;
691 buf[19] = addr & 0xff;
692 addr >>= 8;
693 buf[18] = addr & 0xff;
694 addr >>= 8;
695 buf[17] = addr & 0xff;
696 addr >>= 8;
697 buf[16] = addr & 0x0f;
698 }
699
ip_ipgre_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)700 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
701 {
702 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
703 memcpy(buf, broadcast, 4);
704 else
705 memcpy(buf, &naddr, sizeof(naddr));
706 }
707
708 #if IS_ENABLED(CONFIG_IPV6)
709 #include <linux/ipv6.h>
710 #endif
711
inet_reset_saddr(struct sock * sk)712 static __inline__ void inet_reset_saddr(struct sock *sk)
713 {
714 inet_sk(sk)->inet_saddr = 0;
715 WRITE_ONCE(inet_sk(sk)->inet_rcv_saddr, 0);
716 #if IS_ENABLED(CONFIG_IPV6)
717 if (sk->sk_family == PF_INET6) {
718 struct ipv6_pinfo *np = inet6_sk(sk);
719
720 memset(&np->saddr, 0, sizeof(np->saddr));
721 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
722 }
723 #endif
724 }
725
726 #endif
727
ipv4_addr_hash(__be32 ip)728 static inline unsigned int ipv4_addr_hash(__be32 ip)
729 {
730 return (__force unsigned int) ip;
731 }
732
__ipv4_addr_hash(const __be32 ip,const u32 initval)733 static inline u32 __ipv4_addr_hash(const __be32 ip, const u32 initval)
734 {
735 return jhash_1word((__force u32)ip, initval);
736 }
737
ipv4_portaddr_hash(const struct net * net,__be32 saddr,unsigned int port)738 static inline u32 ipv4_portaddr_hash(const struct net *net,
739 __be32 saddr,
740 unsigned int port)
741 {
742 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
743 }
744
745 bool ip_call_ra_chain(struct sk_buff *skb);
746
747 /*
748 * Functions provided by ip_fragment.c
749 */
750
751 enum ip_defrag_users {
752 IP_DEFRAG_LOCAL_DELIVER,
753 IP_DEFRAG_CALL_RA_CHAIN,
754 IP_DEFRAG_CONNTRACK_IN,
755 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
756 IP_DEFRAG_CONNTRACK_OUT,
757 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
758 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
759 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
760 IP_DEFRAG_VS_IN,
761 IP_DEFRAG_VS_OUT,
762 IP_DEFRAG_VS_FWD,
763 IP_DEFRAG_AF_PACKET,
764 IP_DEFRAG_MACVLAN,
765 };
766
767 /* Return true if the value of 'user' is between 'lower_bond'
768 * and 'upper_bond' inclusively.
769 */
ip_defrag_user_in_between(u32 user,enum ip_defrag_users lower_bond,enum ip_defrag_users upper_bond)770 static inline bool ip_defrag_user_in_between(u32 user,
771 enum ip_defrag_users lower_bond,
772 enum ip_defrag_users upper_bond)
773 {
774 return user >= lower_bond && user <= upper_bond;
775 }
776
777 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
778 #ifdef CONFIG_INET
779 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
780 #else
ip_check_defrag(struct net * net,struct sk_buff * skb,u32 user)781 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
782 {
783 return skb;
784 }
785 #endif
786
787 /*
788 * Functions provided by ip_forward.c
789 */
790
791 int ip_forward(struct sk_buff *skb);
792
793 /*
794 * Functions provided by ip_options.c
795 */
796
797 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
798 __be32 daddr, struct rtable *rt);
799
800 int __ip_options_echo(struct net *net, struct ip_options *dopt,
801 struct sk_buff *skb, const struct ip_options *sopt);
ip_options_echo(struct net * net,struct ip_options * dopt,struct sk_buff * skb)802 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
803 struct sk_buff *skb)
804 {
805 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
806 }
807
808 void ip_options_fragment(struct sk_buff *skb);
809 int __ip_options_compile(struct net *net, struct ip_options *opt,
810 struct sk_buff *skb, __be32 *info);
811 int ip_options_compile(struct net *net, struct ip_options *opt,
812 struct sk_buff *skb);
813 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
814 sockptr_t data, int optlen);
815 void ip_options_undo(struct ip_options *opt);
816 void ip_forward_options(struct sk_buff *skb);
817 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
818
819 /*
820 * Functions provided by ip_sockglue.c
821 */
822
823 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb, bool drop_dst);
824 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
825 struct sk_buff *skb, int tlen, int offset);
826 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
827 struct ipcm_cookie *ipc, bool allow_ipv6);
828 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl);
829 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
830 unsigned int optlen);
831 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
832 unsigned int optlen);
833 int do_ip_getsockopt(struct sock *sk, int level, int optname,
834 sockptr_t optval, sockptr_t optlen);
835 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
836 int __user *optlen);
837 int ip_ra_control(struct sock *sk, unsigned char on,
838 void (*destructor)(struct sock *));
839
840 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len);
841 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
842 u32 info, u8 *payload);
843 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
844 u32 info);
845
ip_cmsg_recv(struct msghdr * msg,struct sk_buff * skb)846 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
847 {
848 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
849 }
850
851 bool icmp_global_allow(struct net *net);
852 void icmp_global_consume(struct net *net);
853
854 #ifdef CONFIG_PROC_FS
855 int ip_misc_proc_init(void);
856 #endif
857
858 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
859 struct netlink_ext_ack *extack);
860
inetdev_valid_mtu(unsigned int mtu)861 static inline bool inetdev_valid_mtu(unsigned int mtu)
862 {
863 return likely(mtu >= IPV4_MIN_MTU);
864 }
865
866 void ip_sock_set_freebind(struct sock *sk);
867 int ip_sock_set_mtu_discover(struct sock *sk, int val);
868 void ip_sock_set_pktinfo(struct sock *sk);
869 void ip_sock_set_recverr(struct sock *sk);
870 void ip_sock_set_tos(struct sock *sk, int val);
871 void __ip_sock_set_tos(struct sock *sk, int val);
872
873 #endif /* _IP_H */
874