• 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 #include <net/lwtunnel.h>
35 
36 #define IPV4_MAX_PMTU		65535U		/* RFC 2675, Section 5.1 */
37 #define IPV4_MIN_MTU		68			/* RFC 791 */
38 
39 extern unsigned int sysctl_fib_sync_mem;
40 extern unsigned int sysctl_fib_sync_mem_min;
41 extern unsigned int sysctl_fib_sync_mem_max;
42 
43 struct sock;
44 
45 struct inet_skb_parm {
46 	int			iif;
47 	struct ip_options	opt;		/* Compiled IP options		*/
48 	u16			flags;
49 
50 #define IPSKB_FORWARDED		BIT(0)
51 #define IPSKB_XFRM_TUNNEL_SIZE	BIT(1)
52 #define IPSKB_XFRM_TRANSFORMED	BIT(2)
53 #define IPSKB_FRAG_COMPLETE	BIT(3)
54 #define IPSKB_REROUTED		BIT(4)
55 #define IPSKB_DOREDIRECT	BIT(5)
56 #define IPSKB_FRAG_PMTU		BIT(6)
57 #define IPSKB_L3SLAVE		BIT(7)
58 #define IPSKB_NOPOLICY		BIT(8)
59 #define IPSKB_MULTIPATH		BIT(9)
60 
61 	u16			frag_max_size;
62 };
63 
ipv4_l3mdev_skb(u16 flags)64 static inline bool ipv4_l3mdev_skb(u16 flags)
65 {
66 	return !!(flags & IPSKB_L3SLAVE);
67 }
68 
ip_hdrlen(const struct sk_buff * skb)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 
ipcm_init(struct ipcm_cookie * ipcm)86 static inline void ipcm_init(struct ipcm_cookie *ipcm)
87 {
88 	*ipcm = (struct ipcm_cookie) { .tos = -1 };
89 }
90 
ipcm_init_sk(struct ipcm_cookie * ipcm,const struct inet_sock * inet)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 = 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 */
inet_sdif(const struct sk_buff * skb)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 
ip_fraglist_next(struct ip_fraglist_iter * iter)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 
ip_finish_skb(struct sock * sk,struct flowi4 * fl4)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 
get_rttos(struct ipcm_cookie * ipc,struct inet_sock * inet)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 
get_rtconn_flags(struct ipcm_cookie * ipc,struct sock * sk)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 
ip_reply_arg_flowi_flags(const struct ip_reply_arg * arg)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 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
299 unsigned long snmp_fold_field(void __percpu *mib, int offt);
300 #if BITS_PER_LONG==32
301 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
302 			 size_t syncp_offset);
303 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
304 #else
snmp_get_cpu_field64(void __percpu * mib,int cpu,int offct,size_t syncp_offset)305 static inline u64  snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
306 					size_t syncp_offset)
307 {
308 	return snmp_get_cpu_field(mib, cpu, offct);
309 
310 }
311 
snmp_fold_field64(void __percpu * mib,int offt,size_t syncp_off)312 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
313 {
314 	return snmp_fold_field(mib, offt);
315 }
316 #endif
317 
318 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
319 { \
320 	int i, c; \
321 	for_each_possible_cpu(c) { \
322 		for (i = 0; stats_list[i].name; i++) \
323 			buff64[i] += snmp_get_cpu_field64( \
324 					mib_statistic, \
325 					c, stats_list[i].entry, \
326 					offset); \
327 	} \
328 }
329 
330 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
331 { \
332 	int i, c; \
333 	for_each_possible_cpu(c) { \
334 		for (i = 0; stats_list[i].name; i++) \
335 			buff[i] += snmp_get_cpu_field( \
336 						mib_statistic, \
337 						c, stats_list[i].entry); \
338 	} \
339 }
340 
341 void inet_get_local_port_range(struct net *net, int *low, int *high);
342 
343 #ifdef CONFIG_SYSCTL
inet_is_local_reserved_port(struct net * net,unsigned short port)344 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
345 {
346 	if (!net->ipv4.sysctl_local_reserved_ports)
347 		return false;
348 	return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
349 }
350 
inet_is_local_unbindable_port(struct net * net,unsigned short port)351 static inline bool inet_is_local_unbindable_port(struct net *net, unsigned short port)
352 {
353 	if (!net->ipv4.sysctl_local_unbindable_ports)
354 		return false;
355 	return test_bit(port, net->ipv4.sysctl_local_unbindable_ports);
356 }
357 
sysctl_dev_name_is_allowed(const char * name)358 static inline bool sysctl_dev_name_is_allowed(const char *name)
359 {
360 	return strcmp(name, "default") != 0  && strcmp(name, "all") != 0;
361 }
362 
inet_port_requires_bind_service(struct net * net,unsigned short port)363 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
364 {
365 	return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
366 }
367 
368 #else
inet_is_local_reserved_port(struct net * net,unsigned short port)369 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
370 {
371 	return false;
372 }
373 
inet_is_local_unbindable_port(struct net * net,unsigned short port)374 static inline bool inet_is_local_unbindable_port(struct net *net, unsigned short port)
375 {
376 	return false;
377 }
378 
inet_port_requires_bind_service(struct net * net,unsigned short port)379 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
380 {
381 	return port < PROT_SOCK;
382 }
383 #endif
384 
385 __be32 inet_current_timestamp(void);
386 
387 /* From inetpeer.c */
388 extern int inet_peer_threshold;
389 extern int inet_peer_minttl;
390 extern int inet_peer_maxttl;
391 
392 void ipfrag_init(void);
393 
394 void ip_static_sysctl_init(void);
395 
396 #define IP4_REPLY_MARK(net, mark) \
397 	(READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
398 
ip_is_fragment(const struct iphdr * iph)399 static inline bool ip_is_fragment(const struct iphdr *iph)
400 {
401 	return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
402 }
403 
404 #ifdef CONFIG_INET
405 #include <net/dst.h>
406 
407 /* The function in 2.2 was invalid, producing wrong result for
408  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
409 static inline
ip_decrease_ttl(struct iphdr * iph)410 int ip_decrease_ttl(struct iphdr *iph)
411 {
412 	u32 check = (__force u32)iph->check;
413 	check += (__force u32)htons(0x0100);
414 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
415 	return --iph->ttl;
416 }
417 
ip_mtu_locked(const struct dst_entry * dst)418 static inline int ip_mtu_locked(const struct dst_entry *dst)
419 {
420 	const struct rtable *rt = (const struct rtable *)dst;
421 
422 	return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
423 }
424 
425 static inline
ip_dont_fragment(const struct sock * sk,const struct dst_entry * dst)426 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
427 {
428 	u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
429 
430 	return  pmtudisc == IP_PMTUDISC_DO ||
431 		(pmtudisc == IP_PMTUDISC_WANT &&
432 		 !ip_mtu_locked(dst));
433 }
434 
ip_sk_accept_pmtu(const struct sock * sk)435 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
436 {
437 	return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
438 	       inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
439 }
440 
ip_sk_use_pmtu(const struct sock * sk)441 static inline bool ip_sk_use_pmtu(const struct sock *sk)
442 {
443 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
444 }
445 
ip_sk_ignore_df(const struct sock * sk)446 static inline bool ip_sk_ignore_df(const struct sock *sk)
447 {
448 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
449 	       inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
450 }
451 
ip_dst_mtu_maybe_forward(const struct dst_entry * dst,bool forwarding)452 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
453 						    bool forwarding)
454 {
455 	const struct rtable *rt = container_of(dst, struct rtable, dst);
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 		mtu = rt->rt_pmtu;
463 		if (mtu && time_before(jiffies, rt->dst.expires))
464 			goto out;
465 	}
466 
467 	/* 'forwarding = true' case should always honour route mtu */
468 	mtu = dst_metric_raw(dst, RTAX_MTU);
469 	if (mtu)
470 		goto out;
471 
472 	mtu = READ_ONCE(dst->dev->mtu);
473 
474 	if (unlikely(ip_mtu_locked(dst))) {
475 		if (rt->rt_uses_gateway && mtu > 576)
476 			mtu = 576;
477 	}
478 
479 out:
480 	mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
481 
482 	return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
483 }
484 
ip_skb_dst_mtu(struct sock * sk,const struct sk_buff * skb)485 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
486 					  const struct sk_buff *skb)
487 {
488 	unsigned int mtu;
489 
490 	if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
491 		bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
492 
493 		return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
494 	}
495 
496 	mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
497 	return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu);
498 }
499 
500 struct dst_metrics *ip_fib_metrics_init(struct net *net, struct nlattr *fc_mx,
501 					int fc_mx_len,
502 					struct netlink_ext_ack *extack);
ip_fib_metrics_put(struct dst_metrics * fib_metrics)503 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
504 {
505 	if (fib_metrics != &dst_default_metrics &&
506 	    refcount_dec_and_test(&fib_metrics->refcnt))
507 		kfree(fib_metrics);
508 }
509 
510 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
511 static inline
ip_dst_init_metrics(struct dst_entry * dst,struct dst_metrics * fib_metrics)512 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
513 {
514 	dst_init_metrics(dst, fib_metrics->metrics, true);
515 
516 	if (fib_metrics != &dst_default_metrics) {
517 		dst->_metrics |= DST_METRICS_REFCOUNTED;
518 		refcount_inc(&fib_metrics->refcnt);
519 	}
520 }
521 
522 static inline
ip_dst_metrics_put(struct dst_entry * dst)523 void ip_dst_metrics_put(struct dst_entry *dst)
524 {
525 	struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
526 
527 	if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
528 		kfree(p);
529 }
530 
531 u32 ip_idents_reserve(u32 hash, int segs);
532 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
533 
ip_select_ident_segs(struct net * net,struct sk_buff * skb,struct sock * sk,int segs)534 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
535 					struct sock *sk, int segs)
536 {
537 	struct iphdr *iph = ip_hdr(skb);
538 
539 	/* We had many attacks based on IPID, use the private
540 	 * generator as much as we can.
541 	 */
542 	if (sk && inet_sk(sk)->inet_daddr) {
543 		iph->id = htons(inet_sk(sk)->inet_id);
544 		inet_sk(sk)->inet_id += segs;
545 		return;
546 	}
547 	if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
548 		iph->id = 0;
549 	} else {
550 		/* Unfortunately we need the big hammer to get a suitable IPID */
551 		__ip_select_ident(net, iph, segs);
552 	}
553 }
554 
ip_select_ident(struct net * net,struct sk_buff * skb,struct sock * sk)555 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
556 				   struct sock *sk)
557 {
558 	ip_select_ident_segs(net, skb, sk, 1);
559 }
560 
inet_compute_pseudo(struct sk_buff * skb,int proto)561 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
562 {
563 	return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
564 				  skb->len, proto, 0);
565 }
566 
567 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
568  * Equivalent to :	flow->v4addrs.src = iph->saddr;
569  *			flow->v4addrs.dst = iph->daddr;
570  */
iph_to_flow_copy_v4addrs(struct flow_keys * flow,const struct iphdr * iph)571 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
572 					    const struct iphdr *iph)
573 {
574 	BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
575 		     offsetof(typeof(flow->addrs), v4addrs.src) +
576 			      sizeof(flow->addrs.v4addrs.src));
577 	memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs));
578 	flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
579 }
580 
inet_gro_compute_pseudo(struct sk_buff * skb,int proto)581 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
582 {
583 	const struct iphdr *iph = skb_gro_network_header(skb);
584 
585 	return csum_tcpudp_nofold(iph->saddr, iph->daddr,
586 				  skb_gro_len(skb), proto, 0);
587 }
588 
589 /*
590  *	Map a multicast IP onto multicast MAC for type ethernet.
591  */
592 
ip_eth_mc_map(__be32 naddr,char * buf)593 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
594 {
595 	__u32 addr=ntohl(naddr);
596 	buf[0]=0x01;
597 	buf[1]=0x00;
598 	buf[2]=0x5e;
599 	buf[5]=addr&0xFF;
600 	addr>>=8;
601 	buf[4]=addr&0xFF;
602 	addr>>=8;
603 	buf[3]=addr&0x7F;
604 }
605 
606 /*
607  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
608  *	Leave P_Key as 0 to be filled in by driver.
609  */
610 
ip_ib_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)611 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
612 {
613 	__u32 addr;
614 	unsigned char scope = broadcast[5] & 0xF;
615 
616 	buf[0]  = 0;		/* Reserved */
617 	buf[1]  = 0xff;		/* Multicast QPN */
618 	buf[2]  = 0xff;
619 	buf[3]  = 0xff;
620 	addr    = ntohl(naddr);
621 	buf[4]  = 0xff;
622 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
623 	buf[6]  = 0x40;		/* IPv4 signature */
624 	buf[7]  = 0x1b;
625 	buf[8]  = broadcast[8];		/* P_Key */
626 	buf[9]  = broadcast[9];
627 	buf[10] = 0;
628 	buf[11] = 0;
629 	buf[12] = 0;
630 	buf[13] = 0;
631 	buf[14] = 0;
632 	buf[15] = 0;
633 	buf[19] = addr & 0xff;
634 	addr  >>= 8;
635 	buf[18] = addr & 0xff;
636 	addr  >>= 8;
637 	buf[17] = addr & 0xff;
638 	addr  >>= 8;
639 	buf[16] = addr & 0x0f;
640 }
641 
ip_ipgre_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)642 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
643 {
644 	if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
645 		memcpy(buf, broadcast, 4);
646 	else
647 		memcpy(buf, &naddr, sizeof(naddr));
648 }
649 
650 #if IS_ENABLED(CONFIG_IPV6)
651 #include <linux/ipv6.h>
652 #endif
653 
inet_reset_saddr(struct sock * sk)654 static __inline__ void inet_reset_saddr(struct sock *sk)
655 {
656 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
657 #if IS_ENABLED(CONFIG_IPV6)
658 	if (sk->sk_family == PF_INET6) {
659 		struct ipv6_pinfo *np = inet6_sk(sk);
660 
661 		memset(&np->saddr, 0, sizeof(np->saddr));
662 		memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
663 	}
664 #endif
665 }
666 
667 #endif
668 
ipv4_addr_hash(__be32 ip)669 static inline unsigned int ipv4_addr_hash(__be32 ip)
670 {
671 	return (__force unsigned int) ip;
672 }
673 
ipv4_portaddr_hash(const struct net * net,__be32 saddr,unsigned int port)674 static inline u32 ipv4_portaddr_hash(const struct net *net,
675 				     __be32 saddr,
676 				     unsigned int port)
677 {
678 	return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
679 }
680 
681 bool ip_call_ra_chain(struct sk_buff *skb);
682 
683 /*
684  *	Functions provided by ip_fragment.c
685  */
686 
687 enum ip_defrag_users {
688 	IP_DEFRAG_LOCAL_DELIVER,
689 	IP_DEFRAG_CALL_RA_CHAIN,
690 	IP_DEFRAG_CONNTRACK_IN,
691 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
692 	IP_DEFRAG_CONNTRACK_OUT,
693 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
694 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
695 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
696 	IP_DEFRAG_VS_IN,
697 	IP_DEFRAG_VS_OUT,
698 	IP_DEFRAG_VS_FWD,
699 	IP_DEFRAG_AF_PACKET,
700 	IP_DEFRAG_MACVLAN,
701 };
702 
703 /* Return true if the value of 'user' is between 'lower_bond'
704  * and 'upper_bond' inclusively.
705  */
ip_defrag_user_in_between(u32 user,enum ip_defrag_users lower_bond,enum ip_defrag_users upper_bond)706 static inline bool ip_defrag_user_in_between(u32 user,
707 					     enum ip_defrag_users lower_bond,
708 					     enum ip_defrag_users upper_bond)
709 {
710 	return user >= lower_bond && user <= upper_bond;
711 }
712 
713 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
714 #ifdef CONFIG_INET
715 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
716 #else
ip_check_defrag(struct net * net,struct sk_buff * skb,u32 user)717 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
718 {
719 	return skb;
720 }
721 #endif
722 
723 /*
724  *	Functions provided by ip_forward.c
725  */
726 
727 int ip_forward(struct sk_buff *skb);
728 
729 /*
730  *	Functions provided by ip_options.c
731  */
732 
733 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
734 		      __be32 daddr, struct rtable *rt, int is_frag);
735 
736 int __ip_options_echo(struct net *net, struct ip_options *dopt,
737 		      struct sk_buff *skb, const struct ip_options *sopt);
ip_options_echo(struct net * net,struct ip_options * dopt,struct sk_buff * skb)738 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
739 				  struct sk_buff *skb)
740 {
741 	return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
742 }
743 
744 void ip_options_fragment(struct sk_buff *skb);
745 int __ip_options_compile(struct net *net, struct ip_options *opt,
746 			 struct sk_buff *skb, __be32 *info);
747 int ip_options_compile(struct net *net, struct ip_options *opt,
748 		       struct sk_buff *skb);
749 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
750 		   sockptr_t data, int optlen);
751 void ip_options_undo(struct ip_options *opt);
752 void ip_forward_options(struct sk_buff *skb);
753 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
754 
755 /*
756  *	Functions provided by ip_sockglue.c
757  */
758 
759 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
760 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
761 			 struct sk_buff *skb, int tlen, int offset);
762 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
763 		 struct ipcm_cookie *ipc, bool allow_ipv6);
764 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
765 		  unsigned int optlen);
766 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
767 		  int __user *optlen);
768 int ip_ra_control(struct sock *sk, unsigned char on,
769 		  void (*destructor)(struct sock *));
770 
771 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
772 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
773 		   u32 info, u8 *payload);
774 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
775 		    u32 info);
776 
ip_cmsg_recv(struct msghdr * msg,struct sk_buff * skb)777 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
778 {
779 	ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
780 }
781 
782 bool icmp_global_allow(void);
783 extern int sysctl_icmp_msgs_per_sec;
784 extern int sysctl_icmp_msgs_burst;
785 
786 #ifdef CONFIG_PROC_FS
787 int ip_misc_proc_init(void);
788 #endif
789 
790 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
791 				struct netlink_ext_ack *extack);
792 
inetdev_valid_mtu(unsigned int mtu)793 static inline bool inetdev_valid_mtu(unsigned int mtu)
794 {
795 	return likely(mtu >= IPV4_MIN_MTU);
796 }
797 
798 void ip_sock_set_freebind(struct sock *sk);
799 int ip_sock_set_mtu_discover(struct sock *sk, int val);
800 void ip_sock_set_pktinfo(struct sock *sk);
801 void ip_sock_set_recverr(struct sock *sk);
802 void ip_sock_set_tos(struct sock *sk, int val);
803 
804 #endif	/* _IP_H */
805