xref: /linux-6.15/include/net/ip.h (revision c4125400)
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, <[email protected]>
12  *		Alan Cox, <[email protected]>
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/snmp.h>
32 #include <net/flow.h>
33 
34 struct sock;
35 
36 struct inet_skb_parm {
37 	struct ip_options	opt;		/* Compiled IP options		*/
38 	unsigned char		flags;
39 
40 #define IPSKB_FORWARDED		1
41 #define IPSKB_XFRM_TUNNEL_SIZE	2
42 #define IPSKB_XFRM_TRANSFORMED	4
43 #define IPSKB_FRAG_COMPLETE	8
44 #define IPSKB_REROUTED		16
45 };
46 
47 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
48 {
49 	return ip_hdr(skb)->ihl * 4;
50 }
51 
52 struct ipcm_cookie {
53 	__be32			addr;
54 	int			oif;
55 	struct ip_options_rcu	*opt;
56 	__u8			tx_flags;
57 };
58 
59 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
60 
61 struct ip_ra_chain {
62 	struct ip_ra_chain __rcu *next;
63 	struct sock		*sk;
64 	union {
65 		void			(*destructor)(struct sock *);
66 		struct sock		*saved_sk;
67 	};
68 	struct rcu_head		rcu;
69 };
70 
71 extern struct ip_ra_chain __rcu *ip_ra_chain;
72 
73 /* IP flags. */
74 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
75 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
76 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
77 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
78 
79 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
80 
81 struct msghdr;
82 struct net_device;
83 struct packet_type;
84 struct rtable;
85 struct sockaddr;
86 
87 extern int		igmp_mc_proc_init(void);
88 
89 /*
90  *	Functions provided by ip.c
91  */
92 
93 extern int		ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
94 					      __be32 saddr, __be32 daddr,
95 					      struct ip_options_rcu *opt);
96 extern int		ip_rcv(struct sk_buff *skb, struct net_device *dev,
97 			       struct packet_type *pt, struct net_device *orig_dev);
98 extern int		ip_local_deliver(struct sk_buff *skb);
99 extern int		ip_mr_input(struct sk_buff *skb);
100 extern int		ip_output(struct sk_buff *skb);
101 extern int		ip_mc_output(struct sk_buff *skb);
102 extern int		ip_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
103 extern int		ip_do_nat(struct sk_buff *skb);
104 extern void		ip_send_check(struct iphdr *ip);
105 extern int		__ip_local_out(struct sk_buff *skb);
106 extern int		ip_local_out(struct sk_buff *skb);
107 extern int		ip_queue_xmit(struct sk_buff *skb, struct flowi *fl);
108 extern void		ip_init(void);
109 extern int		ip_append_data(struct sock *sk, struct flowi4 *fl4,
110 				       int getfrag(void *from, char *to, int offset, int len,
111 						   int odd, struct sk_buff *skb),
112 				void *from, int len, int protolen,
113 				struct ipcm_cookie *ipc,
114 				struct rtable **rt,
115 				unsigned int flags);
116 extern int		ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, struct sk_buff *skb);
117 extern ssize_t		ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
118 				int offset, size_t size, int flags);
119 extern struct sk_buff  *__ip_make_skb(struct sock *sk,
120 				      struct flowi4 *fl4,
121 				      struct sk_buff_head *queue,
122 				      struct inet_cork *cork);
123 extern int		ip_send_skb(struct sk_buff *skb);
124 extern int		ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
125 extern void		ip_flush_pending_frames(struct sock *sk);
126 extern struct sk_buff  *ip_make_skb(struct sock *sk,
127 				    struct flowi4 *fl4,
128 				    int getfrag(void *from, char *to, int offset, int len,
129 						int odd, struct sk_buff *skb),
130 				    void *from, int length, int transhdrlen,
131 				    struct ipcm_cookie *ipc,
132 				    struct rtable **rtp,
133 				    unsigned int flags);
134 
135 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
136 {
137 	return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
138 }
139 
140 /* datagram.c */
141 extern int		ip4_datagram_connect(struct sock *sk,
142 					     struct sockaddr *uaddr, int addr_len);
143 
144 struct ip_reply_arg {
145 	struct kvec iov[1];
146 	int	    flags;
147 	__wsum 	    csum;
148 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
149 				/* -1 if not needed */
150 	int	    bound_dev_if;
151 	u8  	    tos;
152 };
153 
154 #define IP_REPLY_ARG_NOSRCCHECK 1
155 
156 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
157 {
158 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
159 }
160 
161 void ip_send_reply(struct sock *sk, struct sk_buff *skb, __be32 daddr,
162 		   const struct ip_reply_arg *arg, unsigned int len);
163 
164 struct ipv4_config {
165 	int	log_martians;
166 	int	no_pmtu_disc;
167 };
168 
169 extern struct ipv4_config ipv4_config;
170 #define IP_INC_STATS(net, field)	SNMP_INC_STATS64((net)->mib.ip_statistics, field)
171 #define IP_INC_STATS_BH(net, field)	SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
172 #define IP_ADD_STATS(net, field, val)	SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
173 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
174 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
175 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
176 #define NET_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.net_statistics, field)
177 #define NET_INC_STATS_BH(net, field)	SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
178 #define NET_INC_STATS_USER(net, field) 	SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
179 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
180 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
181 
182 extern unsigned long snmp_fold_field(void __percpu *mib[], int offt);
183 #if BITS_PER_LONG==32
184 extern u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t sync_off);
185 #else
186 static inline u64 snmp_fold_field64(void __percpu *mib[], int offt, size_t syncp_off)
187 {
188 	return snmp_fold_field(mib, offt);
189 }
190 #endif
191 extern int snmp_mib_init(void __percpu *ptr[2], size_t mibsize, size_t align);
192 extern void snmp_mib_free(void __percpu *ptr[2]);
193 
194 extern struct local_ports {
195 	seqlock_t	lock;
196 	int		range[2];
197 } sysctl_local_ports;
198 extern void inet_get_local_port_range(int *low, int *high);
199 
200 extern unsigned long *sysctl_local_reserved_ports;
201 static inline int inet_is_reserved_local_port(int port)
202 {
203 	return test_bit(port, sysctl_local_reserved_ports);
204 }
205 
206 extern int sysctl_ip_nonlocal_bind;
207 
208 /* From inetpeer.c */
209 extern int inet_peer_threshold;
210 extern int inet_peer_minttl;
211 extern int inet_peer_maxttl;
212 
213 /* From ip_input.c */
214 extern int sysctl_ip_early_demux;
215 
216 /* From ip_output.c */
217 extern int sysctl_ip_dynaddr;
218 
219 extern void ipfrag_init(void);
220 
221 extern void ip_static_sysctl_init(void);
222 
223 static inline bool ip_is_fragment(const struct iphdr *iph)
224 {
225 	return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
226 }
227 
228 #ifdef CONFIG_INET
229 #include <net/dst.h>
230 
231 /* The function in 2.2 was invalid, producing wrong result for
232  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
233 static inline
234 int ip_decrease_ttl(struct iphdr *iph)
235 {
236 	u32 check = (__force u32)iph->check;
237 	check += (__force u32)htons(0x0100);
238 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
239 	return --iph->ttl;
240 }
241 
242 static inline
243 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
244 {
245 	return  inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
246 		(inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
247 		 !(dst_metric_locked(dst, RTAX_MTU)));
248 }
249 
250 extern void __ip_select_ident(struct iphdr *iph, struct dst_entry *dst, int more);
251 
252 static inline void ip_select_ident(struct iphdr *iph, struct dst_entry *dst, struct sock *sk)
253 {
254 	if (iph->frag_off & htons(IP_DF)) {
255 		/* This is only to work around buggy Windows95/2000
256 		 * VJ compression implementations.  If the ID field
257 		 * does not change, they drop every other packet in
258 		 * a TCP stream using header compression.
259 		 */
260 		iph->id = (sk && inet_sk(sk)->inet_daddr) ?
261 					htons(inet_sk(sk)->inet_id++) : 0;
262 	} else
263 		__ip_select_ident(iph, dst, 0);
264 }
265 
266 static inline void ip_select_ident_more(struct iphdr *iph, struct dst_entry *dst, struct sock *sk, int more)
267 {
268 	if (iph->frag_off & htons(IP_DF)) {
269 		if (sk && inet_sk(sk)->inet_daddr) {
270 			iph->id = htons(inet_sk(sk)->inet_id);
271 			inet_sk(sk)->inet_id += 1 + more;
272 		} else
273 			iph->id = 0;
274 	} else
275 		__ip_select_ident(iph, dst, more);
276 }
277 
278 /*
279  *	Map a multicast IP onto multicast MAC for type ethernet.
280  */
281 
282 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
283 {
284 	__u32 addr=ntohl(naddr);
285 	buf[0]=0x01;
286 	buf[1]=0x00;
287 	buf[2]=0x5e;
288 	buf[5]=addr&0xFF;
289 	addr>>=8;
290 	buf[4]=addr&0xFF;
291 	addr>>=8;
292 	buf[3]=addr&0x7F;
293 }
294 
295 /*
296  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
297  *	Leave P_Key as 0 to be filled in by driver.
298  */
299 
300 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
301 {
302 	__u32 addr;
303 	unsigned char scope = broadcast[5] & 0xF;
304 
305 	buf[0]  = 0;		/* Reserved */
306 	buf[1]  = 0xff;		/* Multicast QPN */
307 	buf[2]  = 0xff;
308 	buf[3]  = 0xff;
309 	addr    = ntohl(naddr);
310 	buf[4]  = 0xff;
311 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
312 	buf[6]  = 0x40;		/* IPv4 signature */
313 	buf[7]  = 0x1b;
314 	buf[8]  = broadcast[8];		/* P_Key */
315 	buf[9]  = broadcast[9];
316 	buf[10] = 0;
317 	buf[11] = 0;
318 	buf[12] = 0;
319 	buf[13] = 0;
320 	buf[14] = 0;
321 	buf[15] = 0;
322 	buf[19] = addr & 0xff;
323 	addr  >>= 8;
324 	buf[18] = addr & 0xff;
325 	addr  >>= 8;
326 	buf[17] = addr & 0xff;
327 	addr  >>= 8;
328 	buf[16] = addr & 0x0f;
329 }
330 
331 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
332 {
333 	if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
334 		memcpy(buf, broadcast, 4);
335 	else
336 		memcpy(buf, &naddr, sizeof(naddr));
337 }
338 
339 #if IS_ENABLED(CONFIG_IPV6)
340 #include <linux/ipv6.h>
341 #endif
342 
343 static __inline__ void inet_reset_saddr(struct sock *sk)
344 {
345 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
346 #if IS_ENABLED(CONFIG_IPV6)
347 	if (sk->sk_family == PF_INET6) {
348 		struct ipv6_pinfo *np = inet6_sk(sk);
349 
350 		memset(&np->saddr, 0, sizeof(np->saddr));
351 		memset(&np->rcv_saddr, 0, sizeof(np->rcv_saddr));
352 	}
353 #endif
354 }
355 
356 #endif
357 
358 static inline int sk_mc_loop(struct sock *sk)
359 {
360 	if (!sk)
361 		return 1;
362 	switch (sk->sk_family) {
363 	case AF_INET:
364 		return inet_sk(sk)->mc_loop;
365 #if IS_ENABLED(CONFIG_IPV6)
366 	case AF_INET6:
367 		return inet6_sk(sk)->mc_loop;
368 #endif
369 	}
370 	WARN_ON(1);
371 	return 1;
372 }
373 
374 extern bool ip_call_ra_chain(struct sk_buff *skb);
375 
376 /*
377  *	Functions provided by ip_fragment.c
378  */
379 
380 enum ip_defrag_users {
381 	IP_DEFRAG_LOCAL_DELIVER,
382 	IP_DEFRAG_CALL_RA_CHAIN,
383 	IP_DEFRAG_CONNTRACK_IN,
384 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
385 	IP_DEFRAG_CONNTRACK_OUT,
386 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
387 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
388 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
389 	IP_DEFRAG_VS_IN,
390 	IP_DEFRAG_VS_OUT,
391 	IP_DEFRAG_VS_FWD,
392 	IP_DEFRAG_AF_PACKET,
393 	IP_DEFRAG_MACVLAN,
394 };
395 
396 int ip_defrag(struct sk_buff *skb, u32 user);
397 #ifdef CONFIG_INET
398 struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user);
399 #else
400 static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user)
401 {
402 	return skb;
403 }
404 #endif
405 int ip_frag_mem(struct net *net);
406 int ip_frag_nqueues(struct net *net);
407 
408 /*
409  *	Functions provided by ip_forward.c
410  */
411 
412 extern int ip_forward(struct sk_buff *skb);
413 
414 /*
415  *	Functions provided by ip_options.c
416  */
417 
418 extern void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
419 			     __be32 daddr, struct rtable *rt, int is_frag);
420 extern int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb);
421 extern void ip_options_fragment(struct sk_buff *skb);
422 extern int ip_options_compile(struct net *net,
423 			      struct ip_options *opt, struct sk_buff *skb);
424 extern int ip_options_get(struct net *net, struct ip_options_rcu **optp,
425 			  unsigned char *data, int optlen);
426 extern int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
427 				    unsigned char __user *data, int optlen);
428 extern void ip_options_undo(struct ip_options * opt);
429 extern void ip_forward_options(struct sk_buff *skb);
430 extern int ip_options_rcv_srr(struct sk_buff *skb);
431 
432 /*
433  *	Functions provided by ip_sockglue.c
434  */
435 
436 extern void	ipv4_pktinfo_prepare(struct sk_buff *skb);
437 extern void	ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb);
438 extern int	ip_cmsg_send(struct net *net,
439 			     struct msghdr *msg, struct ipcm_cookie *ipc);
440 extern int	ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, unsigned int optlen);
441 extern int	ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, int __user *optlen);
442 extern int	compat_ip_setsockopt(struct sock *sk, int level,
443 			int optname, char __user *optval, unsigned int optlen);
444 extern int	compat_ip_getsockopt(struct sock *sk, int level,
445 			int optname, char __user *optval, int __user *optlen);
446 extern int	ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *));
447 
448 extern int 	ip_recv_error(struct sock *sk, struct msghdr *msg, int len);
449 extern void	ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
450 			      __be16 port, u32 info, u8 *payload);
451 extern void	ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
452 			       u32 info);
453 
454 #ifdef CONFIG_PROC_FS
455 extern int ip_misc_proc_init(void);
456 #endif
457 
458 #endif	/* _IP_H */
459