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