xref: /linux/include/net/ip.h (revision 827634added7f38b7d724cab1dccdb2b004c13c3)
1 /*
2  * INET		An implementation of the TCP/IP protocol suite for the LINUX
3  *		operating system.  INET is implemented using the  BSD Socket
4  *		interface as the means of communication with the user level.
5  *
6  *		Definitions for the IP module.
7  *
8  * Version:	@(#)ip.h	1.0.2	05/07/93
9  *
10  * Authors:	Ross Biro
11  *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12  *		Alan Cox, <gw4pts@gw4pts.ampr.org>
13  *
14  * Changes:
15  *		Mike McLagan    :       Routing by source
16  *
17  *		This program is free software; you can redistribute it and/or
18  *		modify it under the terms of the GNU General Public License
19  *		as published by the Free Software Foundation; either version
20  *		2 of the License, or (at your option) any later version.
21  */
22 #ifndef _IP_H
23 #define _IP_H
24 
25 #include <linux/types.h>
26 #include <linux/ip.h>
27 #include <linux/in.h>
28 #include <linux/skbuff.h>
29 
30 #include <net/inet_sock.h>
31 #include <net/route.h>
32 #include <net/snmp.h>
33 #include <net/flow.h>
34 #include <net/flow_keys.h>
35 
36 struct sock;
37 
38 struct inet_skb_parm {
39 	struct ip_options	opt;		/* Compiled IP options		*/
40 	unsigned char		flags;
41 
42 #define IPSKB_FORWARDED		BIT(0)
43 #define IPSKB_XFRM_TUNNEL_SIZE	BIT(1)
44 #define IPSKB_XFRM_TRANSFORMED	BIT(2)
45 #define IPSKB_FRAG_COMPLETE	BIT(3)
46 #define IPSKB_REROUTED		BIT(4)
47 #define IPSKB_DOREDIRECT	BIT(5)
48 
49 	u16			frag_max_size;
50 };
51 
52 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
53 {
54 	return ip_hdr(skb)->ihl * 4;
55 }
56 
57 struct ipcm_cookie {
58 	__be32			addr;
59 	int			oif;
60 	struct ip_options_rcu	*opt;
61 	__u8			tx_flags;
62 	__u8			ttl;
63 	__s16			tos;
64 	char			priority;
65 };
66 
67 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
68 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
69 
70 struct ip_ra_chain {
71 	struct ip_ra_chain __rcu *next;
72 	struct sock		*sk;
73 	union {
74 		void			(*destructor)(struct sock *);
75 		struct sock		*saved_sk;
76 	};
77 	struct rcu_head		rcu;
78 };
79 
80 extern struct ip_ra_chain __rcu *ip_ra_chain;
81 
82 /* IP flags. */
83 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
84 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
85 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
86 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
87 
88 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
89 
90 struct msghdr;
91 struct net_device;
92 struct packet_type;
93 struct rtable;
94 struct sockaddr;
95 
96 int igmp_mc_init(void);
97 
98 /*
99  *	Functions provided by ip.c
100  */
101 
102 int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
103 			  __be32 saddr, __be32 daddr,
104 			  struct ip_options_rcu *opt);
105 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
106 	   struct net_device *orig_dev);
107 int ip_local_deliver(struct sk_buff *skb);
108 int ip_mr_input(struct sk_buff *skb);
109 int ip_output(struct sock *sk, struct sk_buff *skb);
110 int ip_mc_output(struct sock *sk, struct sk_buff *skb);
111 int ip_fragment(struct sock *sk, struct sk_buff *skb,
112 		int (*output)(struct sock *, struct sk_buff *));
113 int ip_do_nat(struct sk_buff *skb);
114 void ip_send_check(struct iphdr *ip);
115 int __ip_local_out(struct sk_buff *skb);
116 int ip_local_out_sk(struct sock *sk, struct sk_buff *skb);
117 static inline int ip_local_out(struct sk_buff *skb)
118 {
119 	return ip_local_out_sk(skb->sk, skb);
120 }
121 
122 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
123 void ip_init(void);
124 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
125 		   int getfrag(void *from, char *to, int offset, int len,
126 			       int odd, struct sk_buff *skb),
127 		   void *from, int len, int protolen,
128 		   struct ipcm_cookie *ipc,
129 		   struct rtable **rt,
130 		   unsigned int flags);
131 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
132 		       struct sk_buff *skb);
133 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
134 		       int offset, size_t size, int flags);
135 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
136 			      struct sk_buff_head *queue,
137 			      struct inet_cork *cork);
138 int ip_send_skb(struct net *net, struct sk_buff *skb);
139 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
140 void ip_flush_pending_frames(struct sock *sk);
141 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
142 			    int getfrag(void *from, char *to, int offset,
143 					int len, int odd, struct sk_buff *skb),
144 			    void *from, int length, int transhdrlen,
145 			    struct ipcm_cookie *ipc, struct rtable **rtp,
146 			    unsigned int flags);
147 
148 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
149 {
150 	return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
151 }
152 
153 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
154 {
155 	return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
156 }
157 
158 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
159 {
160 	return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
161 }
162 
163 /* datagram.c */
164 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
165 
166 void ip4_datagram_release_cb(struct sock *sk);
167 
168 struct ip_reply_arg {
169 	struct kvec iov[1];
170 	int	    flags;
171 	__wsum 	    csum;
172 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
173 				/* -1 if not needed */
174 	int	    bound_dev_if;
175 	u8  	    tos;
176 };
177 
178 #define IP_REPLY_ARG_NOSRCCHECK 1
179 
180 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
181 {
182 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
183 }
184 
185 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
186 			   const struct ip_options *sopt,
187 			   __be32 daddr, __be32 saddr,
188 			   const struct ip_reply_arg *arg,
189 			   unsigned int len);
190 
191 #define IP_INC_STATS(net, field)	SNMP_INC_STATS64((net)->mib.ip_statistics, field)
192 #define IP_INC_STATS_BH(net, field)	SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
193 #define IP_ADD_STATS(net, field, val)	SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
194 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
195 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
196 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
197 #define NET_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.net_statistics, field)
198 #define NET_INC_STATS_BH(net, field)	SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
199 #define NET_INC_STATS_USER(net, field) 	SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
200 #define NET_ADD_STATS(net, field, adnd)	SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
201 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
202 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
203 
204 unsigned long snmp_fold_field(void __percpu *mib, int offt);
205 #if BITS_PER_LONG==32
206 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
207 #else
208 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
209 {
210 	return snmp_fold_field(mib, offt);
211 }
212 #endif
213 
214 void inet_get_local_port_range(struct net *net, int *low, int *high);
215 
216 #ifdef CONFIG_SYSCTL
217 static inline int inet_is_local_reserved_port(struct net *net, int port)
218 {
219 	if (!net->ipv4.sysctl_local_reserved_ports)
220 		return 0;
221 	return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
222 }
223 
224 static inline bool sysctl_dev_name_is_allowed(const char *name)
225 {
226 	return strcmp(name, "default") != 0  && strcmp(name, "all") != 0;
227 }
228 
229 #else
230 static inline int inet_is_local_reserved_port(struct net *net, int port)
231 {
232 	return 0;
233 }
234 #endif
235 
236 /* From inetpeer.c */
237 extern int inet_peer_threshold;
238 extern int inet_peer_minttl;
239 extern int inet_peer_maxttl;
240 
241 /* From ip_input.c */
242 extern int sysctl_ip_early_demux;
243 
244 /* From ip_output.c */
245 extern int sysctl_ip_dynaddr;
246 
247 void ipfrag_init(void);
248 
249 void ip_static_sysctl_init(void);
250 
251 #define IP4_REPLY_MARK(net, mark) \
252 	((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
253 
254 static inline bool ip_is_fragment(const struct iphdr *iph)
255 {
256 	return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
257 }
258 
259 #ifdef CONFIG_INET
260 #include <net/dst.h>
261 
262 /* The function in 2.2 was invalid, producing wrong result for
263  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
264 static inline
265 int ip_decrease_ttl(struct iphdr *iph)
266 {
267 	u32 check = (__force u32)iph->check;
268 	check += (__force u32)htons(0x0100);
269 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
270 	return --iph->ttl;
271 }
272 
273 static inline
274 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
275 {
276 	return  inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
277 		(inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
278 		 !(dst_metric_locked(dst, RTAX_MTU)));
279 }
280 
281 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
282 {
283 	return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
284 	       inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
285 }
286 
287 static inline bool ip_sk_use_pmtu(const struct sock *sk)
288 {
289 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
290 }
291 
292 static inline bool ip_sk_ignore_df(const struct sock *sk)
293 {
294 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
295 	       inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
296 }
297 
298 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
299 						    bool forwarding)
300 {
301 	struct net *net = dev_net(dst->dev);
302 
303 	if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
304 	    dst_metric_locked(dst, RTAX_MTU) ||
305 	    !forwarding)
306 		return dst_mtu(dst);
307 
308 	return min(dst->dev->mtu, IP_MAX_MTU);
309 }
310 
311 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff *skb)
312 {
313 	if (!skb->sk || ip_sk_use_pmtu(skb->sk)) {
314 		bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
315 		return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
316 	} else {
317 		return min(skb_dst(skb)->dev->mtu, IP_MAX_MTU);
318 	}
319 }
320 
321 u32 ip_idents_reserve(u32 hash, int segs);
322 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
323 
324 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
325 					struct sock *sk, int segs)
326 {
327 	struct iphdr *iph = ip_hdr(skb);
328 
329 	if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
330 		/* This is only to work around buggy Windows95/2000
331 		 * VJ compression implementations.  If the ID field
332 		 * does not change, they drop every other packet in
333 		 * a TCP stream using header compression.
334 		 */
335 		if (sk && inet_sk(sk)->inet_daddr) {
336 			iph->id = htons(inet_sk(sk)->inet_id);
337 			inet_sk(sk)->inet_id += segs;
338 		} else {
339 			iph->id = 0;
340 		}
341 	} else {
342 		__ip_select_ident(net, iph, segs);
343 	}
344 }
345 
346 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
347 				   struct sock *sk)
348 {
349 	ip_select_ident_segs(net, skb, sk, 1);
350 }
351 
352 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
353 {
354 	return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
355 				  skb->len, proto, 0);
356 }
357 
358 static inline void inet_set_txhash(struct sock *sk)
359 {
360 	struct inet_sock *inet = inet_sk(sk);
361 	struct flow_keys keys;
362 
363 	keys.src = inet->inet_saddr;
364 	keys.dst = inet->inet_daddr;
365 	keys.port16[0] = inet->inet_sport;
366 	keys.port16[1] = inet->inet_dport;
367 
368 	sk->sk_txhash = flow_hash_from_keys(&keys);
369 }
370 
371 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
372 {
373 	const struct iphdr *iph = skb_gro_network_header(skb);
374 
375 	return csum_tcpudp_nofold(iph->saddr, iph->daddr,
376 				  skb_gro_len(skb), proto, 0);
377 }
378 
379 /*
380  *	Map a multicast IP onto multicast MAC for type ethernet.
381  */
382 
383 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
384 {
385 	__u32 addr=ntohl(naddr);
386 	buf[0]=0x01;
387 	buf[1]=0x00;
388 	buf[2]=0x5e;
389 	buf[5]=addr&0xFF;
390 	addr>>=8;
391 	buf[4]=addr&0xFF;
392 	addr>>=8;
393 	buf[3]=addr&0x7F;
394 }
395 
396 /*
397  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
398  *	Leave P_Key as 0 to be filled in by driver.
399  */
400 
401 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
402 {
403 	__u32 addr;
404 	unsigned char scope = broadcast[5] & 0xF;
405 
406 	buf[0]  = 0;		/* Reserved */
407 	buf[1]  = 0xff;		/* Multicast QPN */
408 	buf[2]  = 0xff;
409 	buf[3]  = 0xff;
410 	addr    = ntohl(naddr);
411 	buf[4]  = 0xff;
412 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
413 	buf[6]  = 0x40;		/* IPv4 signature */
414 	buf[7]  = 0x1b;
415 	buf[8]  = broadcast[8];		/* P_Key */
416 	buf[9]  = broadcast[9];
417 	buf[10] = 0;
418 	buf[11] = 0;
419 	buf[12] = 0;
420 	buf[13] = 0;
421 	buf[14] = 0;
422 	buf[15] = 0;
423 	buf[19] = addr & 0xff;
424 	addr  >>= 8;
425 	buf[18] = addr & 0xff;
426 	addr  >>= 8;
427 	buf[17] = addr & 0xff;
428 	addr  >>= 8;
429 	buf[16] = addr & 0x0f;
430 }
431 
432 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
433 {
434 	if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
435 		memcpy(buf, broadcast, 4);
436 	else
437 		memcpy(buf, &naddr, sizeof(naddr));
438 }
439 
440 #if IS_ENABLED(CONFIG_IPV6)
441 #include <linux/ipv6.h>
442 #endif
443 
444 static __inline__ void inet_reset_saddr(struct sock *sk)
445 {
446 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
447 #if IS_ENABLED(CONFIG_IPV6)
448 	if (sk->sk_family == PF_INET6) {
449 		struct ipv6_pinfo *np = inet6_sk(sk);
450 
451 		memset(&np->saddr, 0, sizeof(np->saddr));
452 		memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
453 	}
454 #endif
455 }
456 
457 #endif
458 
459 bool ip_call_ra_chain(struct sk_buff *skb);
460 
461 /*
462  *	Functions provided by ip_fragment.c
463  */
464 
465 enum ip_defrag_users {
466 	IP_DEFRAG_LOCAL_DELIVER,
467 	IP_DEFRAG_CALL_RA_CHAIN,
468 	IP_DEFRAG_CONNTRACK_IN,
469 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
470 	IP_DEFRAG_CONNTRACK_OUT,
471 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
472 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
473 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
474 	IP_DEFRAG_VS_IN,
475 	IP_DEFRAG_VS_OUT,
476 	IP_DEFRAG_VS_FWD,
477 	IP_DEFRAG_AF_PACKET,
478 	IP_DEFRAG_MACVLAN,
479 };
480 
481 int ip_defrag(struct sk_buff *skb, u32 user);
482 #ifdef CONFIG_INET
483 struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user);
484 #else
485 static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user)
486 {
487 	return skb;
488 }
489 #endif
490 int ip_frag_mem(struct net *net);
491 
492 /*
493  *	Functions provided by ip_forward.c
494  */
495 
496 int ip_forward(struct sk_buff *skb);
497 
498 /*
499  *	Functions provided by ip_options.c
500  */
501 
502 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
503 		      __be32 daddr, struct rtable *rt, int is_frag);
504 
505 int __ip_options_echo(struct ip_options *dopt, struct sk_buff *skb,
506 		      const struct ip_options *sopt);
507 static inline int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb)
508 {
509 	return __ip_options_echo(dopt, skb, &IPCB(skb)->opt);
510 }
511 
512 void ip_options_fragment(struct sk_buff *skb);
513 int ip_options_compile(struct net *net, struct ip_options *opt,
514 		       struct sk_buff *skb);
515 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
516 		   unsigned char *data, int optlen);
517 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
518 			     unsigned char __user *data, int optlen);
519 void ip_options_undo(struct ip_options *opt);
520 void ip_forward_options(struct sk_buff *skb);
521 int ip_options_rcv_srr(struct sk_buff *skb);
522 
523 /*
524  *	Functions provided by ip_sockglue.c
525  */
526 
527 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
528 void ip_cmsg_recv_offset(struct msghdr *msg, struct sk_buff *skb, int offset);
529 int ip_cmsg_send(struct net *net, struct msghdr *msg,
530 		 struct ipcm_cookie *ipc, bool allow_ipv6);
531 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
532 		  unsigned int optlen);
533 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
534 		  int __user *optlen);
535 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
536 			 char __user *optval, unsigned int optlen);
537 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
538 			 char __user *optval, int __user *optlen);
539 int ip_ra_control(struct sock *sk, unsigned char on,
540 		  void (*destructor)(struct sock *));
541 
542 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
543 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
544 		   u32 info, u8 *payload);
545 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
546 		    u32 info);
547 
548 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
549 {
550 	ip_cmsg_recv_offset(msg, skb, 0);
551 }
552 
553 bool icmp_global_allow(void);
554 extern int sysctl_icmp_msgs_per_sec;
555 extern int sysctl_icmp_msgs_burst;
556 
557 #ifdef CONFIG_PROC_FS
558 int ip_misc_proc_init(void);
559 #endif
560 
561 #endif	/* _IP_H */
562