xref: /linux/include/net/ip.h (revision 0526b56cbc3c489642bd6a5fe4b718dea7ef0ee8)
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 
37 #define IPV4_MAX_PMTU		65535U		/* RFC 2675, Section 5.1 */
38 #define IPV4_MIN_MTU		68			/* RFC 791 */
39 
40 extern unsigned int sysctl_fib_sync_mem;
41 extern unsigned int sysctl_fib_sync_mem_min;
42 extern unsigned int sysctl_fib_sync_mem_max;
43 
44 struct sock;
45 
46 struct inet_skb_parm {
47 	int			iif;
48 	struct ip_options	opt;		/* Compiled IP options		*/
49 	u16			flags;
50 
51 #define IPSKB_FORWARDED		BIT(0)
52 #define IPSKB_XFRM_TUNNEL_SIZE	BIT(1)
53 #define IPSKB_XFRM_TRANSFORMED	BIT(2)
54 #define IPSKB_FRAG_COMPLETE	BIT(3)
55 #define IPSKB_REROUTED		BIT(4)
56 #define IPSKB_DOREDIRECT	BIT(5)
57 #define IPSKB_FRAG_PMTU		BIT(6)
58 #define IPSKB_L3SLAVE		BIT(7)
59 #define IPSKB_NOPOLICY		BIT(8)
60 
61 	u16			frag_max_size;
62 };
63 
64 static inline bool ipv4_l3mdev_skb(u16 flags)
65 {
66 	return !!(flags & IPSKB_L3SLAVE);
67 }
68 
69 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
70 {
71 	return ip_hdr(skb)->ihl * 4;
72 }
73 
74 struct ipcm_cookie {
75 	struct sockcm_cookie	sockc;
76 	__be32			addr;
77 	int			oif;
78 	struct ip_options_rcu	*opt;
79 	__u8			protocol;
80 	__u8			ttl;
81 	__s16			tos;
82 	char			priority;
83 	__u16			gso_size;
84 };
85 
86 static inline void ipcm_init(struct ipcm_cookie *ipcm)
87 {
88 	*ipcm = (struct ipcm_cookie) { .tos = -1 };
89 }
90 
91 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
92 				const struct inet_sock *inet)
93 {
94 	ipcm_init(ipcm);
95 
96 	ipcm->sockc.mark = inet->sk.sk_mark;
97 	ipcm->sockc.tsflags = inet->sk.sk_tsflags;
98 	ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if);
99 	ipcm->addr = inet->inet_saddr;
100 	ipcm->protocol = inet->inet_num;
101 }
102 
103 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
104 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
105 
106 /* return enslaved device index if relevant */
107 static inline int inet_sdif(const struct sk_buff *skb)
108 {
109 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
110 	if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
111 		return IPCB(skb)->iif;
112 #endif
113 	return 0;
114 }
115 
116 /* Special input handler for packets caught by router alert option.
117    They are selected only by protocol field, and then processed likely
118    local ones; but only if someone wants them! Otherwise, router
119    not running rsvpd will kill RSVP.
120 
121    It is user level problem, what it will make with them.
122    I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
123    but receiver should be enough clever f.e. to forward mtrace requests,
124    sent to multicast group to reach destination designated router.
125  */
126 
127 struct ip_ra_chain {
128 	struct ip_ra_chain __rcu *next;
129 	struct sock		*sk;
130 	union {
131 		void			(*destructor)(struct sock *);
132 		struct sock		*saved_sk;
133 	};
134 	struct rcu_head		rcu;
135 };
136 
137 /* IP flags. */
138 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
139 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
140 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
141 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
142 
143 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
144 
145 struct msghdr;
146 struct net_device;
147 struct packet_type;
148 struct rtable;
149 struct sockaddr;
150 
151 int igmp_mc_init(void);
152 
153 /*
154  *	Functions provided by ip.c
155  */
156 
157 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
158 			  __be32 saddr, __be32 daddr,
159 			  struct ip_options_rcu *opt, u8 tos);
160 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
161 	   struct net_device *orig_dev);
162 void ip_list_rcv(struct list_head *head, struct packet_type *pt,
163 		 struct net_device *orig_dev);
164 int ip_local_deliver(struct sk_buff *skb);
165 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
166 int ip_mr_input(struct sk_buff *skb);
167 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
168 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
169 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
170 		   int (*output)(struct net *, struct sock *, struct sk_buff *));
171 
172 struct ip_fraglist_iter {
173 	struct sk_buff	*frag;
174 	struct iphdr	*iph;
175 	int		offset;
176 	unsigned int	hlen;
177 };
178 
179 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
180 		      unsigned int hlen, struct ip_fraglist_iter *iter);
181 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
182 
183 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
184 {
185 	struct sk_buff *skb = iter->frag;
186 
187 	iter->frag = skb->next;
188 	skb_mark_not_on_list(skb);
189 
190 	return skb;
191 }
192 
193 struct ip_frag_state {
194 	bool		DF;
195 	unsigned int	hlen;
196 	unsigned int	ll_rs;
197 	unsigned int	mtu;
198 	unsigned int	left;
199 	int		offset;
200 	int		ptr;
201 	__be16		not_last_frag;
202 };
203 
204 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
205 		  unsigned int mtu, bool DF, struct ip_frag_state *state);
206 struct sk_buff *ip_frag_next(struct sk_buff *skb,
207 			     struct ip_frag_state *state);
208 
209 void ip_send_check(struct iphdr *ip);
210 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
211 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
212 
213 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
214 		    __u8 tos);
215 void ip_init(void);
216 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
217 		   int getfrag(void *from, char *to, int offset, int len,
218 			       int odd, struct sk_buff *skb),
219 		   void *from, int len, int protolen,
220 		   struct ipcm_cookie *ipc,
221 		   struct rtable **rt,
222 		   unsigned int flags);
223 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
224 		       struct sk_buff *skb);
225 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
226 		       int offset, size_t size, int flags);
227 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
228 			      struct sk_buff_head *queue,
229 			      struct inet_cork *cork);
230 int ip_send_skb(struct net *net, struct sk_buff *skb);
231 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
232 void ip_flush_pending_frames(struct sock *sk);
233 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
234 			    int getfrag(void *from, char *to, int offset,
235 					int len, int odd, struct sk_buff *skb),
236 			    void *from, int length, int transhdrlen,
237 			    struct ipcm_cookie *ipc, struct rtable **rtp,
238 			    struct inet_cork *cork, unsigned int flags);
239 
240 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
241 
242 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
243 {
244 	return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
245 }
246 
247 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
248 {
249 	return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
250 }
251 
252 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
253 {
254 	return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
255 }
256 
257 /* datagram.c */
258 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
259 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
260 
261 void ip4_datagram_release_cb(struct sock *sk);
262 
263 struct ip_reply_arg {
264 	struct kvec iov[1];
265 	int	    flags;
266 	__wsum 	    csum;
267 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
268 				/* -1 if not needed */
269 	int	    bound_dev_if;
270 	u8  	    tos;
271 	kuid_t	    uid;
272 };
273 
274 #define IP_REPLY_ARG_NOSRCCHECK 1
275 
276 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
277 {
278 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
279 }
280 
281 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
282 			   const struct ip_options *sopt,
283 			   __be32 daddr, __be32 saddr,
284 			   const struct ip_reply_arg *arg,
285 			   unsigned int len, u64 transmit_time);
286 
287 #define IP_INC_STATS(net, field)	SNMP_INC_STATS64((net)->mib.ip_statistics, field)
288 #define __IP_INC_STATS(net, field)	__SNMP_INC_STATS64((net)->mib.ip_statistics, field)
289 #define IP_ADD_STATS(net, field, val)	SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
290 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_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 __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
293 #define NET_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.net_statistics, field)
294 #define __NET_INC_STATS(net, field)	__SNMP_INC_STATS((net)->mib.net_statistics, field)
295 #define NET_ADD_STATS(net, field, adnd)	SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
296 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
297 
298 static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt)
299 {
300 	return  *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt);
301 }
302 
303 unsigned long snmp_fold_field(void __percpu *mib, int offt);
304 #if BITS_PER_LONG==32
305 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
306 			 size_t syncp_offset);
307 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
308 #else
309 static inline u64  snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
310 					size_t syncp_offset)
311 {
312 	return snmp_get_cpu_field(mib, cpu, offct);
313 
314 }
315 
316 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
317 {
318 	return snmp_fold_field(mib, offt);
319 }
320 #endif
321 
322 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
323 { \
324 	int i, c; \
325 	for_each_possible_cpu(c) { \
326 		for (i = 0; stats_list[i].name; i++) \
327 			buff64[i] += snmp_get_cpu_field64( \
328 					mib_statistic, \
329 					c, stats_list[i].entry, \
330 					offset); \
331 	} \
332 }
333 
334 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
335 { \
336 	int i, c; \
337 	for_each_possible_cpu(c) { \
338 		for (i = 0; stats_list[i].name; i++) \
339 			buff[i] += snmp_get_cpu_field( \
340 						mib_statistic, \
341 						c, stats_list[i].entry); \
342 	} \
343 }
344 
345 void inet_get_local_port_range(const struct net *net, int *low, int *high);
346 void inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high);
347 
348 #ifdef CONFIG_SYSCTL
349 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
350 {
351 	if (!net->ipv4.sysctl_local_reserved_ports)
352 		return false;
353 	return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
354 }
355 
356 static inline bool sysctl_dev_name_is_allowed(const char *name)
357 {
358 	return strcmp(name, "default") != 0  && strcmp(name, "all") != 0;
359 }
360 
361 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
362 {
363 	return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
364 }
365 
366 #else
367 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
368 {
369 	return false;
370 }
371 
372 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
373 {
374 	return port < PROT_SOCK;
375 }
376 #endif
377 
378 __be32 inet_current_timestamp(void);
379 
380 /* From inetpeer.c */
381 extern int inet_peer_threshold;
382 extern int inet_peer_minttl;
383 extern int inet_peer_maxttl;
384 
385 void ipfrag_init(void);
386 
387 void ip_static_sysctl_init(void);
388 
389 #define IP4_REPLY_MARK(net, mark) \
390 	(READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
391 
392 static inline bool ip_is_fragment(const struct iphdr *iph)
393 {
394 	return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
395 }
396 
397 #ifdef CONFIG_INET
398 #include <net/dst.h>
399 
400 /* The function in 2.2 was invalid, producing wrong result for
401  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
402 static inline
403 int ip_decrease_ttl(struct iphdr *iph)
404 {
405 	u32 check = (__force u32)iph->check;
406 	check += (__force u32)htons(0x0100);
407 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
408 	return --iph->ttl;
409 }
410 
411 static inline int ip_mtu_locked(const struct dst_entry *dst)
412 {
413 	const struct rtable *rt = (const struct rtable *)dst;
414 
415 	return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
416 }
417 
418 static inline
419 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
420 {
421 	u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
422 
423 	return  pmtudisc == IP_PMTUDISC_DO ||
424 		(pmtudisc == IP_PMTUDISC_WANT &&
425 		 !ip_mtu_locked(dst));
426 }
427 
428 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
429 {
430 	return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
431 	       inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
432 }
433 
434 static inline bool ip_sk_use_pmtu(const struct sock *sk)
435 {
436 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
437 }
438 
439 static inline bool ip_sk_ignore_df(const struct sock *sk)
440 {
441 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
442 	       inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
443 }
444 
445 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
446 						    bool forwarding)
447 {
448 	const struct rtable *rt = container_of(dst, struct rtable, dst);
449 	struct net *net = dev_net(dst->dev);
450 	unsigned int mtu;
451 
452 	if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
453 	    ip_mtu_locked(dst) ||
454 	    !forwarding) {
455 		mtu = rt->rt_pmtu;
456 		if (mtu && time_before(jiffies, rt->dst.expires))
457 			goto out;
458 	}
459 
460 	/* 'forwarding = true' case should always honour route mtu */
461 	mtu = dst_metric_raw(dst, RTAX_MTU);
462 	if (mtu)
463 		goto out;
464 
465 	mtu = READ_ONCE(dst->dev->mtu);
466 
467 	if (unlikely(ip_mtu_locked(dst))) {
468 		if (rt->rt_uses_gateway && mtu > 576)
469 			mtu = 576;
470 	}
471 
472 out:
473 	mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
474 
475 	return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
476 }
477 
478 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
479 					  const struct sk_buff *skb)
480 {
481 	unsigned int mtu;
482 
483 	if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
484 		bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
485 
486 		return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
487 	}
488 
489 	mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
490 	return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu);
491 }
492 
493 struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx,
494 					int fc_mx_len,
495 					struct netlink_ext_ack *extack);
496 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
497 {
498 	if (fib_metrics != &dst_default_metrics &&
499 	    refcount_dec_and_test(&fib_metrics->refcnt))
500 		kfree(fib_metrics);
501 }
502 
503 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
504 static inline
505 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
506 {
507 	dst_init_metrics(dst, fib_metrics->metrics, true);
508 
509 	if (fib_metrics != &dst_default_metrics) {
510 		dst->_metrics |= DST_METRICS_REFCOUNTED;
511 		refcount_inc(&fib_metrics->refcnt);
512 	}
513 }
514 
515 static inline
516 void ip_dst_metrics_put(struct dst_entry *dst)
517 {
518 	struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
519 
520 	if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
521 		kfree(p);
522 }
523 
524 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
525 
526 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
527 					struct sock *sk, int segs)
528 {
529 	struct iphdr *iph = ip_hdr(skb);
530 
531 	/* We had many attacks based on IPID, use the private
532 	 * generator as much as we can.
533 	 */
534 	if (sk && inet_sk(sk)->inet_daddr) {
535 		iph->id = htons(inet_sk(sk)->inet_id);
536 		inet_sk(sk)->inet_id += segs;
537 		return;
538 	}
539 	if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
540 		iph->id = 0;
541 	} else {
542 		/* Unfortunately we need the big hammer to get a suitable IPID */
543 		__ip_select_ident(net, iph, segs);
544 	}
545 }
546 
547 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
548 				   struct sock *sk)
549 {
550 	ip_select_ident_segs(net, skb, sk, 1);
551 }
552 
553 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
554 {
555 	return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
556 				  skb->len, proto, 0);
557 }
558 
559 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
560  * Equivalent to :	flow->v4addrs.src = iph->saddr;
561  *			flow->v4addrs.dst = iph->daddr;
562  */
563 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
564 					    const struct iphdr *iph)
565 {
566 	BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
567 		     offsetof(typeof(flow->addrs), v4addrs.src) +
568 			      sizeof(flow->addrs.v4addrs.src));
569 	memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs));
570 	flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
571 }
572 
573 /*
574  *	Map a multicast IP onto multicast MAC for type ethernet.
575  */
576 
577 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
578 {
579 	__u32 addr=ntohl(naddr);
580 	buf[0]=0x01;
581 	buf[1]=0x00;
582 	buf[2]=0x5e;
583 	buf[5]=addr&0xFF;
584 	addr>>=8;
585 	buf[4]=addr&0xFF;
586 	addr>>=8;
587 	buf[3]=addr&0x7F;
588 }
589 
590 /*
591  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
592  *	Leave P_Key as 0 to be filled in by driver.
593  */
594 
595 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
596 {
597 	__u32 addr;
598 	unsigned char scope = broadcast[5] & 0xF;
599 
600 	buf[0]  = 0;		/* Reserved */
601 	buf[1]  = 0xff;		/* Multicast QPN */
602 	buf[2]  = 0xff;
603 	buf[3]  = 0xff;
604 	addr    = ntohl(naddr);
605 	buf[4]  = 0xff;
606 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
607 	buf[6]  = 0x40;		/* IPv4 signature */
608 	buf[7]  = 0x1b;
609 	buf[8]  = broadcast[8];		/* P_Key */
610 	buf[9]  = broadcast[9];
611 	buf[10] = 0;
612 	buf[11] = 0;
613 	buf[12] = 0;
614 	buf[13] = 0;
615 	buf[14] = 0;
616 	buf[15] = 0;
617 	buf[19] = addr & 0xff;
618 	addr  >>= 8;
619 	buf[18] = addr & 0xff;
620 	addr  >>= 8;
621 	buf[17] = addr & 0xff;
622 	addr  >>= 8;
623 	buf[16] = addr & 0x0f;
624 }
625 
626 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
627 {
628 	if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
629 		memcpy(buf, broadcast, 4);
630 	else
631 		memcpy(buf, &naddr, sizeof(naddr));
632 }
633 
634 #if IS_ENABLED(CONFIG_IPV6)
635 #include <linux/ipv6.h>
636 #endif
637 
638 static __inline__ void inet_reset_saddr(struct sock *sk)
639 {
640 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
641 #if IS_ENABLED(CONFIG_IPV6)
642 	if (sk->sk_family == PF_INET6) {
643 		struct ipv6_pinfo *np = inet6_sk(sk);
644 
645 		memset(&np->saddr, 0, sizeof(np->saddr));
646 		memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
647 	}
648 #endif
649 }
650 
651 #endif
652 
653 static inline unsigned int ipv4_addr_hash(__be32 ip)
654 {
655 	return (__force unsigned int) ip;
656 }
657 
658 static inline u32 ipv4_portaddr_hash(const struct net *net,
659 				     __be32 saddr,
660 				     unsigned int port)
661 {
662 	return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
663 }
664 
665 bool ip_call_ra_chain(struct sk_buff *skb);
666 
667 /*
668  *	Functions provided by ip_fragment.c
669  */
670 
671 enum ip_defrag_users {
672 	IP_DEFRAG_LOCAL_DELIVER,
673 	IP_DEFRAG_CALL_RA_CHAIN,
674 	IP_DEFRAG_CONNTRACK_IN,
675 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
676 	IP_DEFRAG_CONNTRACK_OUT,
677 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
678 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
679 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
680 	IP_DEFRAG_VS_IN,
681 	IP_DEFRAG_VS_OUT,
682 	IP_DEFRAG_VS_FWD,
683 	IP_DEFRAG_AF_PACKET,
684 	IP_DEFRAG_MACVLAN,
685 };
686 
687 /* Return true if the value of 'user' is between 'lower_bond'
688  * and 'upper_bond' inclusively.
689  */
690 static inline bool ip_defrag_user_in_between(u32 user,
691 					     enum ip_defrag_users lower_bond,
692 					     enum ip_defrag_users upper_bond)
693 {
694 	return user >= lower_bond && user <= upper_bond;
695 }
696 
697 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
698 #ifdef CONFIG_INET
699 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
700 #else
701 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
702 {
703 	return skb;
704 }
705 #endif
706 
707 /*
708  *	Functions provided by ip_forward.c
709  */
710 
711 int ip_forward(struct sk_buff *skb);
712 
713 /*
714  *	Functions provided by ip_options.c
715  */
716 
717 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
718 		      __be32 daddr, struct rtable *rt);
719 
720 int __ip_options_echo(struct net *net, struct ip_options *dopt,
721 		      struct sk_buff *skb, const struct ip_options *sopt);
722 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
723 				  struct sk_buff *skb)
724 {
725 	return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
726 }
727 
728 void ip_options_fragment(struct sk_buff *skb);
729 int __ip_options_compile(struct net *net, struct ip_options *opt,
730 			 struct sk_buff *skb, __be32 *info);
731 int ip_options_compile(struct net *net, struct ip_options *opt,
732 		       struct sk_buff *skb);
733 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
734 		   sockptr_t data, int optlen);
735 void ip_options_undo(struct ip_options *opt);
736 void ip_forward_options(struct sk_buff *skb);
737 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
738 
739 /*
740  *	Functions provided by ip_sockglue.c
741  */
742 
743 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
744 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
745 			 struct sk_buff *skb, int tlen, int offset);
746 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
747 		 struct ipcm_cookie *ipc, bool allow_ipv6);
748 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl);
749 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
750 		     unsigned int optlen);
751 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
752 		  unsigned int optlen);
753 int do_ip_getsockopt(struct sock *sk, int level, int optname,
754 		     sockptr_t optval, sockptr_t optlen);
755 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
756 		  int __user *optlen);
757 int ip_ra_control(struct sock *sk, unsigned char on,
758 		  void (*destructor)(struct sock *));
759 
760 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
761 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
762 		   u32 info, u8 *payload);
763 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
764 		    u32 info);
765 
766 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
767 {
768 	ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
769 }
770 
771 bool icmp_global_allow(void);
772 extern int sysctl_icmp_msgs_per_sec;
773 extern int sysctl_icmp_msgs_burst;
774 
775 #ifdef CONFIG_PROC_FS
776 int ip_misc_proc_init(void);
777 #endif
778 
779 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
780 				struct netlink_ext_ack *extack);
781 
782 static inline bool inetdev_valid_mtu(unsigned int mtu)
783 {
784 	return likely(mtu >= IPV4_MIN_MTU);
785 }
786 
787 void ip_sock_set_freebind(struct sock *sk);
788 int ip_sock_set_mtu_discover(struct sock *sk, int val);
789 void ip_sock_set_pktinfo(struct sock *sk);
790 void ip_sock_set_recverr(struct sock *sk);
791 void ip_sock_set_tos(struct sock *sk, int val);
792 void  __ip_sock_set_tos(struct sock *sk, int val);
793 
794 #endif	/* _IP_H */
795