• 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 
27 #include <net/inet_sock.h>
28 #include <net/route.h>
29 #include <net/snmp.h>
30 #include <net/flow.h>
31 #include <net/flow_dissector.h>
32 #include <net/netns/hash.h>
33 #ifndef __GENKSYMS__
34 #include <net/lwtunnel.h>
35 #endif
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 
60 	u16			frag_max_size;
61 };
62 
ipv4_l3mdev_skb(u16 flags)63 static inline bool ipv4_l3mdev_skb(u16 flags)
64 {
65 	return !!(flags & IPSKB_L3SLAVE);
66 }
67 
ip_hdrlen(const struct sk_buff * skb)68 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
69 {
70 	return ip_hdr(skb)->ihl * 4;
71 }
72 
73 struct ipcm_cookie {
74 	struct sockcm_cookie	sockc;
75 	__be32			addr;
76 	int			oif;
77 	struct ip_options_rcu	*opt;
78 	__u8			protocol;
79 	__u8			ttl;
80 	__s16			tos;
81 	char			priority;
82 	__u16			gso_size;
83 };
84 
ipcm_init(struct ipcm_cookie * ipcm)85 static inline void ipcm_init(struct ipcm_cookie *ipcm)
86 {
87 	*ipcm = (struct ipcm_cookie) { .tos = -1 };
88 }
89 
ipcm_init_sk(struct ipcm_cookie * ipcm,const struct inet_sock * inet)90 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
91 				const struct inet_sock *inet)
92 {
93 	ipcm_init(ipcm);
94 
95 	ipcm->sockc.mark = inet->sk.sk_mark;
96 	ipcm->sockc.tsflags = inet->sk.sk_tsflags;
97 	ipcm->oif = inet->sk.sk_bound_dev_if;
98 	ipcm->addr = inet->inet_saddr;
99 	ipcm->protocol = inet->inet_num;
100 }
101 
102 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
103 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
104 
105 /* return enslaved device index if relevant */
inet_sdif(struct sk_buff * skb)106 static inline int inet_sdif(struct sk_buff *skb)
107 {
108 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
109 	if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
110 		return IPCB(skb)->iif;
111 #endif
112 	return 0;
113 }
114 
115 /* Special input handler for packets caught by router alert option.
116    They are selected only by protocol field, and then processed likely
117    local ones; but only if someone wants them! Otherwise, router
118    not running rsvpd will kill RSVP.
119 
120    It is user level problem, what it will make with them.
121    I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
122    but receiver should be enough clever f.e. to forward mtrace requests,
123    sent to multicast group to reach destination designated router.
124  */
125 
126 struct ip_ra_chain {
127 	struct ip_ra_chain __rcu *next;
128 	struct sock		*sk;
129 	union {
130 		void			(*destructor)(struct sock *);
131 		struct sock		*saved_sk;
132 	};
133 	struct rcu_head		rcu;
134 };
135 
136 /* IP flags. */
137 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
138 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
139 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
140 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
141 
142 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
143 
144 struct msghdr;
145 struct net_device;
146 struct packet_type;
147 struct rtable;
148 struct sockaddr;
149 
150 int igmp_mc_init(void);
151 
152 /*
153  *	Functions provided by ip.c
154  */
155 
156 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
157 			  __be32 saddr, __be32 daddr,
158 			  struct ip_options_rcu *opt);
159 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
160 	   struct net_device *orig_dev);
161 void ip_list_rcv(struct list_head *head, struct packet_type *pt,
162 		 struct net_device *orig_dev);
163 int ip_local_deliver(struct sk_buff *skb);
164 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
165 int ip_mr_input(struct sk_buff *skb);
166 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
167 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
168 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
169 		   int (*output)(struct net *, struct sock *, struct sk_buff *));
170 
171 struct ip_fraglist_iter {
172 	struct sk_buff	*frag;
173 	struct iphdr	*iph;
174 	int		offset;
175 	unsigned int	hlen;
176 };
177 
178 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
179 		      unsigned int hlen, struct ip_fraglist_iter *iter);
180 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
181 
ip_fraglist_next(struct ip_fraglist_iter * iter)182 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
183 {
184 	struct sk_buff *skb = iter->frag;
185 
186 	iter->frag = skb->next;
187 	skb_mark_not_on_list(skb);
188 
189 	return skb;
190 }
191 
192 struct ip_frag_state {
193 	bool		DF;
194 	unsigned int	hlen;
195 	unsigned int	ll_rs;
196 	unsigned int	mtu;
197 	unsigned int	left;
198 	int		offset;
199 	int		ptr;
200 	__be16		not_last_frag;
201 };
202 
203 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
204 		  unsigned int mtu, bool DF, struct ip_frag_state *state);
205 struct sk_buff *ip_frag_next(struct sk_buff *skb,
206 			     struct ip_frag_state *state);
207 
208 void ip_send_check(struct iphdr *ip);
209 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
210 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
211 
212 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
213 		    __u8 tos);
214 void ip_init(void);
215 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
216 		   int getfrag(void *from, char *to, int offset, int len,
217 			       int odd, struct sk_buff *skb),
218 		   void *from, int len, int protolen,
219 		   struct ipcm_cookie *ipc,
220 		   struct rtable **rt,
221 		   unsigned int flags);
222 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
223 		       struct sk_buff *skb);
224 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
225 		       int offset, size_t size, int flags);
226 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
227 			      struct sk_buff_head *queue,
228 			      struct inet_cork *cork);
229 int ip_send_skb(struct net *net, struct sk_buff *skb);
230 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
231 void ip_flush_pending_frames(struct sock *sk);
232 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
233 			    int getfrag(void *from, char *to, int offset,
234 					int len, int odd, struct sk_buff *skb),
235 			    void *from, int length, int transhdrlen,
236 			    struct ipcm_cookie *ipc, struct rtable **rtp,
237 			    struct inet_cork *cork, unsigned int flags);
238 
ip_queue_xmit(struct sock * sk,struct sk_buff * skb,struct flowi * fl)239 static inline int ip_queue_xmit(struct sock *sk, struct sk_buff *skb,
240 				struct flowi *fl)
241 {
242 	return __ip_queue_xmit(sk, skb, fl, inet_sk(sk)->tos);
243 }
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 
get_rttos(struct ipcm_cookie * ipc,struct inet_sock * inet)250 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
251 {
252 	return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
253 }
254 
get_rtconn_flags(struct ipcm_cookie * ipc,struct sock * sk)255 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
256 {
257 	return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
258 }
259 
260 /* datagram.c */
261 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
262 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
263 
264 void ip4_datagram_release_cb(struct sock *sk);
265 
266 struct ip_reply_arg {
267 	struct kvec iov[1];
268 	int	    flags;
269 	__wsum 	    csum;
270 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
271 				/* -1 if not needed */
272 	int	    bound_dev_if;
273 	u8  	    tos;
274 	kuid_t	    uid;
275 };
276 
277 #define IP_REPLY_ARG_NOSRCCHECK 1
278 
ip_reply_arg_flowi_flags(const struct ip_reply_arg * arg)279 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
280 {
281 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
282 }
283 
284 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
285 			   const struct ip_options *sopt,
286 			   __be32 daddr, __be32 saddr,
287 			   const struct ip_reply_arg *arg,
288 			   unsigned int len, u64 transmit_time);
289 
290 #define IP_INC_STATS(net, field)	SNMP_INC_STATS64((net)->mib.ip_statistics, field)
291 #define __IP_INC_STATS(net, field)	__SNMP_INC_STATS64((net)->mib.ip_statistics, field)
292 #define IP_ADD_STATS(net, field, val)	SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
293 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
294 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
295 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
296 #define NET_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.net_statistics, field)
297 #define __NET_INC_STATS(net, field)	__SNMP_INC_STATS((net)->mib.net_statistics, field)
298 #define NET_ADD_STATS(net, field, adnd)	SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
299 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
300 
301 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
302 unsigned long snmp_fold_field(void __percpu *mib, int offt);
303 #if BITS_PER_LONG==32
304 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
305 			 size_t syncp_offset);
306 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
307 #else
snmp_get_cpu_field64(void __percpu * mib,int cpu,int offct,size_t syncp_offset)308 static inline u64  snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
309 					size_t syncp_offset)
310 {
311 	return snmp_get_cpu_field(mib, cpu, offct);
312 
313 }
314 
snmp_fold_field64(void __percpu * mib,int offt,size_t syncp_off)315 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
316 {
317 	return snmp_fold_field(mib, offt);
318 }
319 #endif
320 
321 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
322 { \
323 	int i, c; \
324 	for_each_possible_cpu(c) { \
325 		for (i = 0; stats_list[i].name; i++) \
326 			buff64[i] += snmp_get_cpu_field64( \
327 					mib_statistic, \
328 					c, stats_list[i].entry, \
329 					offset); \
330 	} \
331 }
332 
333 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
334 { \
335 	int i, c; \
336 	for_each_possible_cpu(c) { \
337 		for (i = 0; stats_list[i].name; i++) \
338 			buff[i] += snmp_get_cpu_field( \
339 						mib_statistic, \
340 						c, stats_list[i].entry); \
341 	} \
342 }
343 
344 void inet_get_local_port_range(struct net *net, int *low, int *high);
345 
346 #ifdef CONFIG_SYSCTL
inet_is_local_reserved_port(struct net * net,int port)347 static inline int inet_is_local_reserved_port(struct net *net, int port)
348 {
349 	if (!net->ipv4.sysctl_local_reserved_ports)
350 		return 0;
351 	return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
352 }
353 
inet_is_local_unbindable_port(struct net * net,unsigned short port)354 static inline bool inet_is_local_unbindable_port(struct net *net, unsigned short port)
355 {
356 	if (!net->ipv4.sysctl_local_unbindable_ports)
357 		return false;
358 	return test_bit(port, net->ipv4.sysctl_local_unbindable_ports);
359 }
360 
sysctl_dev_name_is_allowed(const char * name)361 static inline bool sysctl_dev_name_is_allowed(const char *name)
362 {
363 	return strcmp(name, "default") != 0  && strcmp(name, "all") != 0;
364 }
365 
inet_prot_sock(struct net * net)366 static inline int inet_prot_sock(struct net *net)
367 {
368 	return net->ipv4.sysctl_ip_prot_sock;
369 }
370 
371 #else
inet_is_local_reserved_port(struct net * net,int port)372 static inline int inet_is_local_reserved_port(struct net *net, int port)
373 {
374 	return 0;
375 }
376 
inet_is_local_unbindable_port(struct net * net,unsigned short port)377 static inline bool inet_is_local_unbindable_port(struct net *net, unsigned short port)
378 {
379 	return false;
380 }
381 
inet_prot_sock(struct net * net)382 static inline int inet_prot_sock(struct net *net)
383 {
384 	return PROT_SOCK;
385 }
386 #endif
387 
388 __be32 inet_current_timestamp(void);
389 
390 /* From inetpeer.c */
391 extern int inet_peer_threshold;
392 extern int inet_peer_minttl;
393 extern int inet_peer_maxttl;
394 
395 void ipfrag_init(void);
396 
397 void ip_static_sysctl_init(void);
398 
399 #define IP4_REPLY_MARK(net, mark) \
400 	(READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
401 
ip_is_fragment(const struct iphdr * iph)402 static inline bool ip_is_fragment(const struct iphdr *iph)
403 {
404 	return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
405 }
406 
407 #ifdef CONFIG_INET
408 #include <net/dst.h>
409 
410 /* The function in 2.2 was invalid, producing wrong result for
411  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
412 static inline
ip_decrease_ttl(struct iphdr * iph)413 int ip_decrease_ttl(struct iphdr *iph)
414 {
415 	u32 check = (__force u32)iph->check;
416 	check += (__force u32)htons(0x0100);
417 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
418 	return --iph->ttl;
419 }
420 
ip_mtu_locked(const struct dst_entry * dst)421 static inline int ip_mtu_locked(const struct dst_entry *dst)
422 {
423 	const struct rtable *rt = (const struct rtable *)dst;
424 
425 	return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
426 }
427 
428 static inline
ip_dont_fragment(const struct sock * sk,const struct dst_entry * dst)429 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
430 {
431 	u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
432 
433 	return  pmtudisc == IP_PMTUDISC_DO ||
434 		(pmtudisc == IP_PMTUDISC_WANT &&
435 		 !ip_mtu_locked(dst));
436 }
437 
ip_sk_accept_pmtu(const struct sock * sk)438 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
439 {
440 	return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
441 	       inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
442 }
443 
ip_sk_use_pmtu(const struct sock * sk)444 static inline bool ip_sk_use_pmtu(const struct sock *sk)
445 {
446 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
447 }
448 
ip_sk_ignore_df(const struct sock * sk)449 static inline bool ip_sk_ignore_df(const struct sock *sk)
450 {
451 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
452 	       inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
453 }
454 
ip_dst_mtu_maybe_forward(const struct dst_entry * dst,bool forwarding)455 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
456 						    bool forwarding)
457 {
458 	struct net *net = dev_net(dst->dev);
459 	unsigned int mtu;
460 
461 	if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
462 	    ip_mtu_locked(dst) ||
463 	    !forwarding)
464 		return dst_mtu(dst);
465 
466 	/* 'forwarding = true' case should always honour route mtu */
467 	mtu = dst_metric_raw(dst, RTAX_MTU);
468 	if (!mtu)
469 		mtu = min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU);
470 
471 	return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
472 }
473 
ip_skb_dst_mtu(struct sock * sk,const struct sk_buff * skb)474 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
475 					  const struct sk_buff *skb)
476 {
477 	unsigned int mtu;
478 
479 	if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
480 		bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
481 
482 		return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
483 	}
484 
485 	mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
486 	return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu);
487 }
488 
489 struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx,
490 					int fc_mx_len,
491 					struct netlink_ext_ack *extack);
ip_fib_metrics_put(struct dst_metrics * fib_metrics)492 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
493 {
494 	if (fib_metrics != &dst_default_metrics &&
495 	    refcount_dec_and_test(&fib_metrics->refcnt))
496 		kfree(fib_metrics);
497 }
498 
499 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
500 static inline
ip_dst_init_metrics(struct dst_entry * dst,struct dst_metrics * fib_metrics)501 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
502 {
503 	dst_init_metrics(dst, fib_metrics->metrics, true);
504 
505 	if (fib_metrics != &dst_default_metrics) {
506 		dst->_metrics |= DST_METRICS_REFCOUNTED;
507 		refcount_inc(&fib_metrics->refcnt);
508 	}
509 }
510 
511 static inline
ip_dst_metrics_put(struct dst_entry * dst)512 void ip_dst_metrics_put(struct dst_entry *dst)
513 {
514 	struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
515 
516 	if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
517 		kfree(p);
518 }
519 
520 u32 ip_idents_reserve(u32 hash, int segs);
521 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
522 
ip_select_ident_segs(struct net * net,struct sk_buff * skb,struct sock * sk,int segs)523 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
524 					struct sock *sk, int segs)
525 {
526 	struct iphdr *iph = ip_hdr(skb);
527 
528 	/* We had many attacks based on IPID, use the private
529 	 * generator as much as we can.
530 	 */
531 	if (sk && inet_sk(sk)->inet_daddr) {
532 		iph->id = htons(inet_sk(sk)->inet_id);
533 		inet_sk(sk)->inet_id += segs;
534 		return;
535 	}
536 	if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
537 		iph->id = 0;
538 	} else {
539 		/* Unfortunately we need the big hammer to get a suitable IPID */
540 		__ip_select_ident(net, iph, segs);
541 	}
542 }
543 
ip_select_ident(struct net * net,struct sk_buff * skb,struct sock * sk)544 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
545 				   struct sock *sk)
546 {
547 	ip_select_ident_segs(net, skb, sk, 1);
548 }
549 
inet_compute_pseudo(struct sk_buff * skb,int proto)550 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
551 {
552 	return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
553 				  skb->len, proto, 0);
554 }
555 
556 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
557  * Equivalent to :	flow->v4addrs.src = iph->saddr;
558  *			flow->v4addrs.dst = iph->daddr;
559  */
iph_to_flow_copy_v4addrs(struct flow_keys * flow,const struct iphdr * iph)560 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
561 					    const struct iphdr *iph)
562 {
563 	BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
564 		     offsetof(typeof(flow->addrs), v4addrs.src) +
565 			      sizeof(flow->addrs.v4addrs.src));
566 	memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs));
567 	flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
568 }
569 
inet_gro_compute_pseudo(struct sk_buff * skb,int proto)570 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
571 {
572 	const struct iphdr *iph = skb_gro_network_header(skb);
573 
574 	return csum_tcpudp_nofold(iph->saddr, iph->daddr,
575 				  skb_gro_len(skb), proto, 0);
576 }
577 
578 /*
579  *	Map a multicast IP onto multicast MAC for type ethernet.
580  */
581 
ip_eth_mc_map(__be32 naddr,char * buf)582 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
583 {
584 	__u32 addr=ntohl(naddr);
585 	buf[0]=0x01;
586 	buf[1]=0x00;
587 	buf[2]=0x5e;
588 	buf[5]=addr&0xFF;
589 	addr>>=8;
590 	buf[4]=addr&0xFF;
591 	addr>>=8;
592 	buf[3]=addr&0x7F;
593 }
594 
595 /*
596  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
597  *	Leave P_Key as 0 to be filled in by driver.
598  */
599 
ip_ib_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)600 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
601 {
602 	__u32 addr;
603 	unsigned char scope = broadcast[5] & 0xF;
604 
605 	buf[0]  = 0;		/* Reserved */
606 	buf[1]  = 0xff;		/* Multicast QPN */
607 	buf[2]  = 0xff;
608 	buf[3]  = 0xff;
609 	addr    = ntohl(naddr);
610 	buf[4]  = 0xff;
611 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
612 	buf[6]  = 0x40;		/* IPv4 signature */
613 	buf[7]  = 0x1b;
614 	buf[8]  = broadcast[8];		/* P_Key */
615 	buf[9]  = broadcast[9];
616 	buf[10] = 0;
617 	buf[11] = 0;
618 	buf[12] = 0;
619 	buf[13] = 0;
620 	buf[14] = 0;
621 	buf[15] = 0;
622 	buf[19] = addr & 0xff;
623 	addr  >>= 8;
624 	buf[18] = addr & 0xff;
625 	addr  >>= 8;
626 	buf[17] = addr & 0xff;
627 	addr  >>= 8;
628 	buf[16] = addr & 0x0f;
629 }
630 
ip_ipgre_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)631 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
632 {
633 	if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
634 		memcpy(buf, broadcast, 4);
635 	else
636 		memcpy(buf, &naddr, sizeof(naddr));
637 }
638 
639 #if IS_ENABLED(CONFIG_IPV6)
640 #include <linux/ipv6.h>
641 #endif
642 
inet_reset_saddr(struct sock * sk)643 static __inline__ void inet_reset_saddr(struct sock *sk)
644 {
645 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
646 #if IS_ENABLED(CONFIG_IPV6)
647 	if (sk->sk_family == PF_INET6) {
648 		struct ipv6_pinfo *np = inet6_sk(sk);
649 
650 		memset(&np->saddr, 0, sizeof(np->saddr));
651 		memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
652 	}
653 #endif
654 }
655 
656 #endif
657 
ipv4_addr_hash(__be32 ip)658 static inline unsigned int ipv4_addr_hash(__be32 ip)
659 {
660 	return (__force unsigned int) ip;
661 }
662 
ipv4_portaddr_hash(const struct net * net,__be32 saddr,unsigned int port)663 static inline u32 ipv4_portaddr_hash(const struct net *net,
664 				     __be32 saddr,
665 				     unsigned int port)
666 {
667 	return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
668 }
669 
670 bool ip_call_ra_chain(struct sk_buff *skb);
671 
672 /*
673  *	Functions provided by ip_fragment.c
674  */
675 
676 enum ip_defrag_users {
677 	IP_DEFRAG_LOCAL_DELIVER,
678 	IP_DEFRAG_CALL_RA_CHAIN,
679 	IP_DEFRAG_CONNTRACK_IN,
680 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
681 	IP_DEFRAG_CONNTRACK_OUT,
682 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
683 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
684 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
685 	IP_DEFRAG_VS_IN,
686 	IP_DEFRAG_VS_OUT,
687 	IP_DEFRAG_VS_FWD,
688 	IP_DEFRAG_AF_PACKET,
689 	IP_DEFRAG_MACVLAN,
690 };
691 
692 /* Return true if the value of 'user' is between 'lower_bond'
693  * and 'upper_bond' inclusively.
694  */
ip_defrag_user_in_between(u32 user,enum ip_defrag_users lower_bond,enum ip_defrag_users upper_bond)695 static inline bool ip_defrag_user_in_between(u32 user,
696 					     enum ip_defrag_users lower_bond,
697 					     enum ip_defrag_users upper_bond)
698 {
699 	return user >= lower_bond && user <= upper_bond;
700 }
701 
702 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
703 #ifdef CONFIG_INET
704 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
705 #else
ip_check_defrag(struct net * net,struct sk_buff * skb,u32 user)706 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
707 {
708 	return skb;
709 }
710 #endif
711 
712 /*
713  *	Functions provided by ip_forward.c
714  */
715 
716 int ip_forward(struct sk_buff *skb);
717 
718 /*
719  *	Functions provided by ip_options.c
720  */
721 
722 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
723 		      __be32 daddr, struct rtable *rt, int is_frag);
724 
725 int __ip_options_echo(struct net *net, struct ip_options *dopt,
726 		      struct sk_buff *skb, const struct ip_options *sopt);
ip_options_echo(struct net * net,struct ip_options * dopt,struct sk_buff * skb)727 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
728 				  struct sk_buff *skb)
729 {
730 	return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
731 }
732 
733 void ip_options_fragment(struct sk_buff *skb);
734 int __ip_options_compile(struct net *net, struct ip_options *opt,
735 			 struct sk_buff *skb, __be32 *info);
736 int ip_options_compile(struct net *net, struct ip_options *opt,
737 		       struct sk_buff *skb);
738 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
739 		   unsigned char *data, int optlen);
740 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
741 			     unsigned char __user *data, int optlen);
742 void ip_options_undo(struct ip_options *opt);
743 void ip_forward_options(struct sk_buff *skb);
744 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
745 
746 /*
747  *	Functions provided by ip_sockglue.c
748  */
749 
750 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
751 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
752 			 struct sk_buff *skb, int tlen, int offset);
753 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
754 		 struct ipcm_cookie *ipc, bool allow_ipv6);
755 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
756 		  unsigned int optlen);
757 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
758 		  int __user *optlen);
759 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
760 			 char __user *optval, unsigned int optlen);
761 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
762 			 char __user *optval, int __user *optlen);
763 int ip_ra_control(struct sock *sk, unsigned char on,
764 		  void (*destructor)(struct sock *));
765 
766 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
767 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
768 		   u32 info, u8 *payload);
769 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
770 		    u32 info);
771 
ip_cmsg_recv(struct msghdr * msg,struct sk_buff * skb)772 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
773 {
774 	ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
775 }
776 
777 bool icmp_global_allow(void);
778 extern int sysctl_icmp_msgs_per_sec;
779 extern int sysctl_icmp_msgs_burst;
780 
781 #ifdef CONFIG_PROC_FS
782 int ip_misc_proc_init(void);
783 #endif
784 
785 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
786 				struct netlink_ext_ack *extack);
787 
inetdev_valid_mtu(unsigned int mtu)788 static inline bool inetdev_valid_mtu(unsigned int mtu)
789 {
790 	return likely(mtu >= IPV4_MIN_MTU);
791 }
792 
793 #endif	/* _IP_H */
794