xref: /openbmc/linux/include/net/ip.h (revision 1fa6ac37)
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 	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	*opt;
56 	union skb_shared_tx	shtx;
57 };
58 
59 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
60 
61 struct ip_ra_chain {
62 	struct ip_ra_chain	*next;
63 	struct sock		*sk;
64 	void			(*destructor)(struct sock *);
65 };
66 
67 extern struct ip_ra_chain *ip_ra_chain;
68 extern rwlock_t ip_ra_lock;
69 
70 /* IP flags. */
71 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
72 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
73 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
74 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
75 
76 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
77 
78 struct msghdr;
79 struct net_device;
80 struct packet_type;
81 struct rtable;
82 struct sockaddr;
83 
84 extern int		igmp_mc_proc_init(void);
85 
86 /*
87  *	Functions provided by ip.c
88  */
89 
90 extern int		ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
91 					      __be32 saddr, __be32 daddr,
92 					      struct ip_options *opt);
93 extern int		ip_rcv(struct sk_buff *skb, struct net_device *dev,
94 			       struct packet_type *pt, struct net_device *orig_dev);
95 extern int		ip_local_deliver(struct sk_buff *skb);
96 extern int		ip_mr_input(struct sk_buff *skb);
97 extern int		ip_output(struct sk_buff *skb);
98 extern int		ip_mc_output(struct sk_buff *skb);
99 extern int		ip_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
100 extern int		ip_do_nat(struct sk_buff *skb);
101 extern void		ip_send_check(struct iphdr *ip);
102 extern int		__ip_local_out(struct sk_buff *skb);
103 extern int		ip_local_out(struct sk_buff *skb);
104 extern int		ip_queue_xmit(struct sk_buff *skb);
105 extern void		ip_init(void);
106 extern int		ip_append_data(struct sock *sk,
107 				       int getfrag(void *from, char *to, int offset, int len,
108 						   int odd, struct sk_buff *skb),
109 				void *from, int len, int protolen,
110 				struct ipcm_cookie *ipc,
111 				struct rtable **rt,
112 				unsigned int flags);
113 extern int		ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, struct sk_buff *skb);
114 extern ssize_t		ip_append_page(struct sock *sk, struct page *page,
115 				int offset, size_t size, int flags);
116 extern int		ip_push_pending_frames(struct sock *sk);
117 extern void		ip_flush_pending_frames(struct sock *sk);
118 
119 /* datagram.c */
120 extern int		ip4_datagram_connect(struct sock *sk,
121 					     struct sockaddr *uaddr, int addr_len);
122 
123 /*
124  *	Map a multicast IP onto multicast MAC for type Token Ring.
125  *      This conforms to RFC1469 Option 2 Multicasting i.e.
126  *      using a functional address to transmit / receive
127  *      multicast packets.
128  */
129 
130 static inline void ip_tr_mc_map(__be32 addr, char *buf)
131 {
132 	buf[0]=0xC0;
133 	buf[1]=0x00;
134 	buf[2]=0x00;
135 	buf[3]=0x04;
136 	buf[4]=0x00;
137 	buf[5]=0x00;
138 }
139 
140 struct ip_reply_arg {
141 	struct kvec iov[1];
142 	int	    flags;
143 	__wsum 	    csum;
144 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
145 				/* -1 if not needed */
146 	int	    bound_dev_if;
147 };
148 
149 #define IP_REPLY_ARG_NOSRCCHECK 1
150 
151 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
152 {
153 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
154 }
155 
156 void ip_send_reply(struct sock *sk, struct sk_buff *skb, struct ip_reply_arg *arg,
157 		   unsigned int len);
158 
159 struct ipv4_config {
160 	int	log_martians;
161 	int	no_pmtu_disc;
162 };
163 
164 extern struct ipv4_config ipv4_config;
165 #define IP_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.ip_statistics, field)
166 #define IP_INC_STATS_BH(net, field)	SNMP_INC_STATS_BH((net)->mib.ip_statistics, field)
167 #define IP_ADD_STATS(net, field, val)	SNMP_ADD_STATS((net)->mib.ip_statistics, field, val)
168 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS_BH((net)->mib.ip_statistics, field, val)
169 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS((net)->mib.ip_statistics, field, val)
170 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_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 __percpu *mib[], int offt);
178 extern int snmp_mib_init(void __percpu *ptr[2], size_t mibsize);
179 extern void snmp_mib_free(void __percpu *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 unsigned long *sysctl_local_reserved_ports;
188 static inline int inet_is_reserved_local_port(int port)
189 {
190 	return test_bit(port, sysctl_local_reserved_ports);
191 }
192 
193 extern int sysctl_ip_default_ttl;
194 extern int sysctl_ip_nonlocal_bind;
195 
196 extern struct ctl_path net_core_path[];
197 extern struct ctl_path net_ipv4_ctl_path[];
198 
199 /* From inetpeer.c */
200 extern int inet_peer_threshold;
201 extern int inet_peer_minttl;
202 extern int inet_peer_maxttl;
203 extern int inet_peer_gc_mintime;
204 extern int inet_peer_gc_maxtime;
205 
206 /* From ip_output.c */
207 extern int sysctl_ip_dynaddr;
208 
209 extern void ipfrag_init(void);
210 
211 extern void ip_static_sysctl_init(void);
212 
213 #ifdef CONFIG_INET
214 #include <net/dst.h>
215 
216 /* The function in 2.2 was invalid, producing wrong result for
217  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
218 static inline
219 int ip_decrease_ttl(struct iphdr *iph)
220 {
221 	u32 check = (__force u32)iph->check;
222 	check += (__force u32)htons(0x0100);
223 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
224 	return --iph->ttl;
225 }
226 
227 static inline
228 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
229 {
230 	return (inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
231 		(inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
232 		 !(dst_metric_locked(dst, RTAX_MTU))));
233 }
234 
235 extern void __ip_select_ident(struct iphdr *iph, struct dst_entry *dst, int more);
236 
237 static inline void ip_select_ident(struct iphdr *iph, struct dst_entry *dst, struct sock *sk)
238 {
239 	if (iph->frag_off & htons(IP_DF)) {
240 		/* This is only to work around buggy Windows95/2000
241 		 * VJ compression implementations.  If the ID field
242 		 * does not change, they drop every other packet in
243 		 * a TCP stream using header compression.
244 		 */
245 		iph->id = (sk && inet_sk(sk)->inet_daddr) ?
246 					htons(inet_sk(sk)->inet_id++) : 0;
247 	} else
248 		__ip_select_ident(iph, dst, 0);
249 }
250 
251 static inline void ip_select_ident_more(struct iphdr *iph, struct dst_entry *dst, struct sock *sk, int more)
252 {
253 	if (iph->frag_off & htons(IP_DF)) {
254 		if (sk && inet_sk(sk)->inet_daddr) {
255 			iph->id = htons(inet_sk(sk)->inet_id);
256 			inet_sk(sk)->inet_id += 1 + more;
257 		} else
258 			iph->id = 0;
259 	} else
260 		__ip_select_ident(iph, dst, more);
261 }
262 
263 /*
264  *	Map a multicast IP onto multicast MAC for type ethernet.
265  */
266 
267 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
268 {
269 	__u32 addr=ntohl(naddr);
270 	buf[0]=0x01;
271 	buf[1]=0x00;
272 	buf[2]=0x5e;
273 	buf[5]=addr&0xFF;
274 	addr>>=8;
275 	buf[4]=addr&0xFF;
276 	addr>>=8;
277 	buf[3]=addr&0x7F;
278 }
279 
280 /*
281  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
282  *	Leave P_Key as 0 to be filled in by driver.
283  */
284 
285 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
286 {
287 	__u32 addr;
288 	unsigned char scope = broadcast[5] & 0xF;
289 
290 	buf[0]  = 0;		/* Reserved */
291 	buf[1]  = 0xff;		/* Multicast QPN */
292 	buf[2]  = 0xff;
293 	buf[3]  = 0xff;
294 	addr    = ntohl(naddr);
295 	buf[4]  = 0xff;
296 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
297 	buf[6]  = 0x40;		/* IPv4 signature */
298 	buf[7]  = 0x1b;
299 	buf[8]  = broadcast[8];		/* P_Key */
300 	buf[9]  = broadcast[9];
301 	buf[10] = 0;
302 	buf[11] = 0;
303 	buf[12] = 0;
304 	buf[13] = 0;
305 	buf[14] = 0;
306 	buf[15] = 0;
307 	buf[19] = addr & 0xff;
308 	addr  >>= 8;
309 	buf[18] = addr & 0xff;
310 	addr  >>= 8;
311 	buf[17] = addr & 0xff;
312 	addr  >>= 8;
313 	buf[16] = addr & 0x0f;
314 }
315 
316 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
317 #include <linux/ipv6.h>
318 #endif
319 
320 static __inline__ void inet_reset_saddr(struct sock *sk)
321 {
322 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
323 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
324 	if (sk->sk_family == PF_INET6) {
325 		struct ipv6_pinfo *np = inet6_sk(sk);
326 
327 		memset(&np->saddr, 0, sizeof(np->saddr));
328 		memset(&np->rcv_saddr, 0, sizeof(np->rcv_saddr));
329 	}
330 #endif
331 }
332 
333 #endif
334 
335 static inline int sk_mc_loop(struct sock *sk)
336 {
337 	if (!sk)
338 		return 1;
339 	switch (sk->sk_family) {
340 	case AF_INET:
341 		return inet_sk(sk)->mc_loop;
342 #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
343 	case AF_INET6:
344 		return inet6_sk(sk)->mc_loop;
345 #endif
346 	}
347 	WARN_ON(1);
348 	return 1;
349 }
350 
351 extern int	ip_call_ra_chain(struct sk_buff *skb);
352 
353 /*
354  *	Functions provided by ip_fragment.c
355  */
356 
357 enum ip_defrag_users {
358 	IP_DEFRAG_LOCAL_DELIVER,
359 	IP_DEFRAG_CALL_RA_CHAIN,
360 	IP_DEFRAG_CONNTRACK_IN,
361 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
362 	IP_DEFRAG_CONNTRACK_OUT,
363 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
364 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
365 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
366 	IP_DEFRAG_VS_IN,
367 	IP_DEFRAG_VS_OUT,
368 	IP_DEFRAG_VS_FWD
369 };
370 
371 int ip_defrag(struct sk_buff *skb, u32 user);
372 int ip_frag_mem(struct net *net);
373 int ip_frag_nqueues(struct net *net);
374 
375 /*
376  *	Functions provided by ip_forward.c
377  */
378 
379 extern int ip_forward(struct sk_buff *skb);
380 
381 /*
382  *	Functions provided by ip_options.c
383  */
384 
385 extern void ip_options_build(struct sk_buff *skb, struct ip_options *opt, __be32 daddr, struct rtable *rt, int is_frag);
386 extern int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb);
387 extern void ip_options_fragment(struct sk_buff *skb);
388 extern int ip_options_compile(struct net *net,
389 			      struct ip_options *opt, struct sk_buff *skb);
390 extern int ip_options_get(struct net *net, struct ip_options **optp,
391 			  unsigned char *data, int optlen);
392 extern int ip_options_get_from_user(struct net *net, struct ip_options **optp,
393 				    unsigned char __user *data, int optlen);
394 extern void ip_options_undo(struct ip_options * opt);
395 extern void ip_forward_options(struct sk_buff *skb);
396 extern int ip_options_rcv_srr(struct sk_buff *skb);
397 
398 /*
399  *	Functions provided by ip_sockglue.c
400  */
401 
402 extern int	ip_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
403 extern void	ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb);
404 extern int	ip_cmsg_send(struct net *net,
405 			     struct msghdr *msg, struct ipcm_cookie *ipc);
406 extern int	ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval, unsigned int optlen);
407 extern int	ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, int __user *optlen);
408 extern int	compat_ip_setsockopt(struct sock *sk, int level,
409 			int optname, char __user *optval, unsigned int optlen);
410 extern int	compat_ip_getsockopt(struct sock *sk, int level,
411 			int optname, char __user *optval, int __user *optlen);
412 extern int	ip_ra_control(struct sock *sk, unsigned char on, void (*destructor)(struct sock *));
413 
414 extern int 	ip_recv_error(struct sock *sk, struct msghdr *msg, int len);
415 extern void	ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err,
416 			      __be16 port, u32 info, u8 *payload);
417 extern void	ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
418 			       u32 info);
419 
420 /* sysctl helpers - any sysctl which holds a value that ends up being
421  * fed into the routing cache should use these handlers.
422  */
423 int ipv4_doint_and_flush(ctl_table *ctl, int write,
424 			 void __user *buffer,
425 			 size_t *lenp, loff_t *ppos);
426 int ipv4_doint_and_flush_strategy(ctl_table *table,
427 				  void __user *oldval, size_t __user *oldlenp,
428 				  void __user *newval, size_t newlen);
429 #ifdef CONFIG_PROC_FS
430 extern int ip_misc_proc_init(void);
431 #endif
432 
433 #endif	/* _IP_H */
434