xref: /linux/include/net/ip.h (revision 367b8112fe2ea5c39a7bb4d263dcdd9b612fae18)
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/snmp.h>
32 #include <net/flow.h>
33 
34 struct sock;
35 
36 struct inet_skb_parm
37 {
38 	struct ip_options	opt;		/* Compiled IP options		*/
39 	unsigned char		flags;
40 
41 #define IPSKB_FORWARDED		1
42 #define IPSKB_XFRM_TUNNEL_SIZE	2
43 #define IPSKB_XFRM_TRANSFORMED	4
44 #define IPSKB_FRAG_COMPLETE	8
45 #define IPSKB_REROUTED		16
46 };
47 
48 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
49 {
50 	return ip_hdr(skb)->ihl * 4;
51 }
52 
53 struct ipcm_cookie
54 {
55 	__be32			addr;
56 	int			oif;
57 	struct ip_options	*opt;
58 };
59 
60 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
61 
62 struct ip_ra_chain
63 {
64 	struct ip_ra_chain	*next;
65 	struct sock		*sk;
66 	void			(*destructor)(struct sock *);
67 };
68 
69 extern struct ip_ra_chain *ip_ra_chain;
70 extern rwlock_t ip_ra_lock;
71 
72 /* IP flags. */
73 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
74 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
75 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
76 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
77 
78 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
79 
80 struct msghdr;
81 struct net_device;
82 struct packet_type;
83 struct rtable;
84 struct sockaddr;
85 
86 extern int		igmp_mc_proc_init(void);
87 
88 /*
89  *	Functions provided by ip.c
90  */
91 
92 extern int		ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
93 					      __be32 saddr, __be32 daddr,
94 					      struct ip_options *opt);
95 extern int		ip_rcv(struct sk_buff *skb, struct net_device *dev,
96 			       struct packet_type *pt, struct net_device *orig_dev);
97 extern int		ip_local_deliver(struct sk_buff *skb);
98 extern int		ip_mr_input(struct sk_buff *skb);
99 extern int		ip_output(struct sk_buff *skb);
100 extern int		ip_mc_output(struct sk_buff *skb);
101 extern int		ip_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
102 extern int		ip_do_nat(struct sk_buff *skb);
103 extern void		ip_send_check(struct iphdr *ip);
104 extern int		__ip_local_out(struct sk_buff *skb);
105 extern int		ip_local_out(struct sk_buff *skb);
106 extern int		ip_queue_xmit(struct sk_buff *skb, int ipfragok);
107 extern void		ip_init(void);
108 extern int		ip_append_data(struct sock *sk,
109 				       int getfrag(void *from, char *to, int offset, int len,
110 						   int odd, struct sk_buff *skb),
111 				void *from, int len, int protolen,
112 				struct ipcm_cookie *ipc,
113 				struct rtable *rt,
114 				unsigned int flags);
115 extern int		ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, struct sk_buff *skb);
116 extern ssize_t		ip_append_page(struct sock *sk, struct page *page,
117 				int offset, size_t size, int flags);
118 extern int		ip_push_pending_frames(struct sock *sk);
119 extern void		ip_flush_pending_frames(struct sock *sk);
120 
121 /* datagram.c */
122 extern int		ip4_datagram_connect(struct sock *sk,
123 					     struct sockaddr *uaddr, int addr_len);
124 
125 /*
126  *	Map a multicast IP onto multicast MAC for type Token Ring.
127  *      This conforms to RFC1469 Option 2 Multicasting i.e.
128  *      using a functional address to transmit / receive
129  *      multicast packets.
130  */
131 
132 static inline void ip_tr_mc_map(__be32 addr, char *buf)
133 {
134 	buf[0]=0xC0;
135 	buf[1]=0x00;
136 	buf[2]=0x00;
137 	buf[3]=0x04;
138 	buf[4]=0x00;
139 	buf[5]=0x00;
140 }
141 
142 struct ip_reply_arg {
143 	struct kvec iov[1];
144 	int	    flags;
145 	__wsum 	    csum;
146 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
147 				/* -1 if not needed */
148 	int	    bound_dev_if;
149 };
150 
151 #define IP_REPLY_ARG_NOSRCCHECK 1
152 
153 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
154 {
155 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
156 }
157 
158 void ip_send_reply(struct sock *sk, struct sk_buff *skb, struct ip_reply_arg *arg,
159 		   unsigned int len);
160 
161 struct ipv4_config
162 {
163 	int	log_martians;
164 	int	no_pmtu_disc;
165 };
166 
167 extern struct ipv4_config ipv4_config;
168 #define IP_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.ip_statistics, field)
169 #define IP_INC_STATS_BH(net, field)	SNMP_INC_STATS_BH((net)->mib.ip_statistics, field)
170 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS_BH((net)->mib.ip_statistics, field, val)
171 #define NET_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.net_statistics, field)
172 #define NET_INC_STATS_BH(net, field)	SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
173 #define NET_INC_STATS_USER(net, field) 	SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
174 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
175 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
176 
177 extern unsigned long snmp_fold_field(void *mib[], int offt);
178 extern int snmp_mib_init(void *ptr[2], size_t mibsize);
179 extern void snmp_mib_free(void *ptr[2]);
180 
181 extern struct local_ports {
182 	seqlock_t	lock;
183 	int		range[2];
184 } sysctl_local_ports;
185 extern void inet_get_local_port_range(int *low, int *high);
186 
187 extern int sysctl_ip_default_ttl;
188 extern int sysctl_ip_nonlocal_bind;
189 
190 extern struct ctl_path net_ipv4_ctl_path[];
191 
192 /* From inetpeer.c */
193 extern int inet_peer_threshold;
194 extern int inet_peer_minttl;
195 extern int inet_peer_maxttl;
196 extern int inet_peer_gc_mintime;
197 extern int inet_peer_gc_maxtime;
198 
199 /* From ip_output.c */
200 extern int sysctl_ip_dynaddr;
201 
202 extern void ipfrag_init(void);
203 
204 extern void ip_static_sysctl_init(void);
205 
206 #ifdef CONFIG_INET
207 #include <net/dst.h>
208 
209 /* The function in 2.2 was invalid, producing wrong result for
210  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
211 static inline
212 int ip_decrease_ttl(struct iphdr *iph)
213 {
214 	u32 check = (__force u32)iph->check;
215 	check += (__force u32)htons(0x0100);
216 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
217 	return --iph->ttl;
218 }
219 
220 static inline
221 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
222 {
223 	return (inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
224 		(inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
225 		 !(dst_metric_locked(dst, RTAX_MTU))));
226 }
227 
228 extern void __ip_select_ident(struct iphdr *iph, struct dst_entry *dst, int more);
229 
230 static inline void ip_select_ident(struct iphdr *iph, struct dst_entry *dst, struct sock *sk)
231 {
232 	if (iph->frag_off & htons(IP_DF)) {
233 		/* This is only to work around buggy Windows95/2000
234 		 * VJ compression implementations.  If the ID field
235 		 * does not change, they drop every other packet in
236 		 * a TCP stream using header compression.
237 		 */
238 		iph->id = (sk && inet_sk(sk)->daddr) ?
239 					htons(inet_sk(sk)->id++) : 0;
240 	} else
241 		__ip_select_ident(iph, dst, 0);
242 }
243 
244 static inline void ip_select_ident_more(struct iphdr *iph, struct dst_entry *dst, struct sock *sk, int more)
245 {
246 	if (iph->frag_off & htons(IP_DF)) {
247 		if (sk && inet_sk(sk)->daddr) {
248 			iph->id = htons(inet_sk(sk)->id);
249 			inet_sk(sk)->id += 1 + more;
250 		} else
251 			iph->id = 0;
252 	} else
253 		__ip_select_ident(iph, dst, more);
254 }
255 
256 /*
257  *	Map a multicast IP onto multicast MAC for type ethernet.
258  */
259 
260 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
261 {
262 	__u32 addr=ntohl(naddr);
263 	buf[0]=0x01;
264 	buf[1]=0x00;
265 	buf[2]=0x5e;
266 	buf[5]=addr&0xFF;
267 	addr>>=8;
268 	buf[4]=addr&0xFF;
269 	addr>>=8;
270 	buf[3]=addr&0x7F;
271 }
272 
273 /*
274  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
275  *	Leave P_Key as 0 to be filled in by driver.
276  */
277 
278 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
279 {
280 	__u32 addr;
281 	unsigned char scope = broadcast[5] & 0xF;
282 
283 	buf[0]  = 0;		/* Reserved */
284 	buf[1]  = 0xff;		/* Multicast QPN */
285 	buf[2]  = 0xff;
286 	buf[3]  = 0xff;
287 	addr    = ntohl(naddr);
288 	buf[4]  = 0xff;
289 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
290 	buf[6]  = 0x40;		/* IPv4 signature */
291 	buf[7]  = 0x1b;
292 	buf[8]  = broadcast[8];		/* P_Key */
293 	buf[9]  = broadcast[9];
294 	buf[10] = 0;
295 	buf[11] = 0;
296 	buf[12] = 0;
297 	buf[13] = 0;
298 	buf[14] = 0;
299 	buf[15] = 0;
300 	buf[19] = addr & 0xff;
301 	addr  >>= 8;
302 	buf[18] = addr & 0xff;
303 	addr  >>= 8;
304 	buf[17] = addr & 0xff;
305 	addr  >>= 8;
306 	buf[16] = addr & 0x0f;
307 }
308 
309 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
310 #include <linux/ipv6.h>
311 #endif
312 
313 static __inline__ void inet_reset_saddr(struct sock *sk)
314 {
315 	inet_sk(sk)->rcv_saddr = inet_sk(sk)->saddr = 0;
316 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
317 	if (sk->sk_family == PF_INET6) {
318 		struct ipv6_pinfo *np = inet6_sk(sk);
319 
320 		memset(&np->saddr, 0, sizeof(np->saddr));
321 		memset(&np->rcv_saddr, 0, sizeof(np->rcv_saddr));
322 	}
323 #endif
324 }
325 
326 #endif
327 
328 extern int	ip_call_ra_chain(struct sk_buff *skb);
329 
330 /*
331  *	Functions provided by ip_fragment.c
332  */
333 
334 enum ip_defrag_users
335 {
336 	IP_DEFRAG_LOCAL_DELIVER,
337 	IP_DEFRAG_CALL_RA_CHAIN,
338 	IP_DEFRAG_CONNTRACK_IN,
339 	IP_DEFRAG_CONNTRACK_OUT,
340 	IP_DEFRAG_VS_IN,
341 	IP_DEFRAG_VS_OUT,
342 	IP_DEFRAG_VS_FWD
343 };
344 
345 int ip_defrag(struct sk_buff *skb, u32 user);
346 int ip_frag_mem(struct net *net);
347 int ip_frag_nqueues(struct net *net);
348 
349 /*
350  *	Functions provided by ip_forward.c
351  */
352 
353 extern int ip_forward(struct sk_buff *skb);
354 
355 /*
356  *	Functions provided by ip_options.c
357  */
358 
359 extern void ip_options_build(struct sk_buff *skb, struct ip_options *opt, __be32 daddr, struct rtable *rt, int is_frag);
360 extern int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb);
361 extern void ip_options_fragment(struct sk_buff *skb);
362 extern int ip_options_compile(struct net *net,
363 			      struct ip_options *opt, struct sk_buff *skb);
364 extern int ip_options_get(struct net *net, struct ip_options **optp,
365 			  unsigned char *data, int optlen);
366 extern int ip_options_get_from_user(struct net *net, struct ip_options **optp,
367 				    unsigned char __user *data, int optlen);
368 extern void ip_options_undo(struct ip_options * opt);
369 extern void ip_forward_options(struct sk_buff *skb);
370 extern int ip_options_rcv_srr(struct sk_buff *skb);
371 
372 /*
373  *	Functions provided by ip_sockglue.c
374  */
375 
376 extern void	ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb);
377 extern int	ip_cmsg_send(struct net *net,
378 			     struct msghdr *msg, struct ipcm_cookie *ipc);
379 extern int	ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, int optlen);
380 extern int	ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, int __user *optlen);
381 extern int	compat_ip_setsockopt(struct sock *sk, int level,
382 			int optname, char __user *optval, int optlen);
383 extern int	compat_ip_getsockopt(struct sock *sk, int level,
384 			int optname, char __user *optval, int __user *optlen);
385 extern int	ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *));
386 
387 extern int 	ip_recv_error(struct sock *sk, struct msghdr *msg, int len);
388 extern void	ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
389 			      __be16 port, u32 info, u8 *payload);
390 extern void	ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
391 			       u32 info);
392 
393 /* sysctl helpers - any sysctl which holds a value that ends up being
394  * fed into the routing cache should use these handlers.
395  */
396 int ipv4_doint_and_flush(ctl_table *ctl, int write,
397 			 struct file* filp, void __user *buffer,
398 			 size_t *lenp, loff_t *ppos);
399 int ipv4_doint_and_flush_strategy(ctl_table *table,
400 				  void __user *oldval, size_t __user *oldlenp,
401 				  void __user *newval, size_t newlen);
402 #ifdef CONFIG_PROC_FS
403 extern int ip_misc_proc_init(void);
404 #endif
405 
406 #endif	/* _IP_H */
407