xref: /openbmc/linux/include/net/ip.h (revision d2999e1b)
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 
35 struct sock;
36 
37 struct inet_skb_parm {
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 	u16			frag_max_size;
48 };
49 
50 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
51 {
52 	return ip_hdr(skb)->ihl * 4;
53 }
54 
55 struct ipcm_cookie {
56 	__be32			addr;
57 	int			oif;
58 	struct ip_options_rcu	*opt;
59 	__u8			tx_flags;
60 	__u8			ttl;
61 	__s16			tos;
62 	char			priority;
63 };
64 
65 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
66 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
67 
68 struct ip_ra_chain {
69 	struct ip_ra_chain __rcu *next;
70 	struct sock		*sk;
71 	union {
72 		void			(*destructor)(struct sock *);
73 		struct sock		*saved_sk;
74 	};
75 	struct rcu_head		rcu;
76 };
77 
78 extern struct ip_ra_chain __rcu *ip_ra_chain;
79 
80 /* IP flags. */
81 #define IP_CE		0x8000		/* Flag: "Congestion"		*/
82 #define IP_DF		0x4000		/* Flag: "Don't Fragment"	*/
83 #define IP_MF		0x2000		/* Flag: "More Fragments"	*/
84 #define IP_OFFSET	0x1FFF		/* "Fragment Offset" part	*/
85 
86 #define IP_FRAG_TIME	(30 * HZ)		/* fragment lifetime	*/
87 
88 struct msghdr;
89 struct net_device;
90 struct packet_type;
91 struct rtable;
92 struct sockaddr;
93 
94 int igmp_mc_init(void);
95 
96 /*
97  *	Functions provided by ip.c
98  */
99 
100 int ip_build_and_send_pkt(struct sk_buff *skb, struct sock *sk,
101 			  __be32 saddr, __be32 daddr,
102 			  struct ip_options_rcu *opt);
103 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
104 	   struct net_device *orig_dev);
105 int ip_local_deliver(struct sk_buff *skb);
106 int ip_mr_input(struct sk_buff *skb);
107 int ip_output(struct sock *sk, struct sk_buff *skb);
108 int ip_mc_output(struct sock *sk, struct sk_buff *skb);
109 int ip_fragment(struct sk_buff *skb, int (*output)(struct sk_buff *));
110 int ip_do_nat(struct sk_buff *skb);
111 void ip_send_check(struct iphdr *ip);
112 int __ip_local_out(struct sk_buff *skb);
113 int ip_local_out_sk(struct sock *sk, struct sk_buff *skb);
114 static inline int ip_local_out(struct sk_buff *skb)
115 {
116 	return ip_local_out_sk(skb->sk, skb);
117 }
118 
119 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
120 void ip_init(void);
121 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
122 		   int getfrag(void *from, char *to, int offset, int len,
123 			       int odd, struct sk_buff *skb),
124 		   void *from, int len, int protolen,
125 		   struct ipcm_cookie *ipc,
126 		   struct rtable **rt,
127 		   unsigned int flags);
128 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
129 		       struct sk_buff *skb);
130 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
131 		       int offset, size_t size, int flags);
132 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
133 			      struct sk_buff_head *queue,
134 			      struct inet_cork *cork);
135 int ip_send_skb(struct net *net, struct sk_buff *skb);
136 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
137 void ip_flush_pending_frames(struct sock *sk);
138 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
139 			    int getfrag(void *from, char *to, int offset,
140 					int len, int odd, struct sk_buff *skb),
141 			    void *from, int length, int transhdrlen,
142 			    struct ipcm_cookie *ipc, struct rtable **rtp,
143 			    unsigned int flags);
144 
145 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
146 {
147 	return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
148 }
149 
150 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
151 {
152 	return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
153 }
154 
155 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
156 {
157 	return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
158 }
159 
160 /* datagram.c */
161 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
162 
163 void ip4_datagram_release_cb(struct sock *sk);
164 
165 struct ip_reply_arg {
166 	struct kvec iov[1];
167 	int	    flags;
168 	__wsum 	    csum;
169 	int	    csumoffset; /* u16 offset of csum in iov[0].iov_base */
170 				/* -1 if not needed */
171 	int	    bound_dev_if;
172 	u8  	    tos;
173 };
174 
175 #define IP_REPLY_ARG_NOSRCCHECK 1
176 
177 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
178 {
179 	return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
180 }
181 
182 void ip_send_unicast_reply(struct net *net, struct sk_buff *skb, __be32 daddr,
183 			   __be32 saddr, const struct ip_reply_arg *arg,
184 			   unsigned int len);
185 
186 #define IP_INC_STATS(net, field)	SNMP_INC_STATS64((net)->mib.ip_statistics, field)
187 #define IP_INC_STATS_BH(net, field)	SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
188 #define IP_ADD_STATS(net, field, val)	SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
189 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
190 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
191 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
192 #define NET_INC_STATS(net, field)	SNMP_INC_STATS((net)->mib.net_statistics, field)
193 #define NET_INC_STATS_BH(net, field)	SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
194 #define NET_INC_STATS_USER(net, field) 	SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
195 #define NET_ADD_STATS(net, field, adnd)	SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
196 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
197 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
198 
199 unsigned long snmp_fold_field(void __percpu *mib, int offt);
200 #if BITS_PER_LONG==32
201 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
202 #else
203 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
204 {
205 	return snmp_fold_field(mib, offt);
206 }
207 #endif
208 
209 void inet_get_local_port_range(struct net *net, int *low, int *high);
210 
211 #ifdef CONFIG_SYSCTL
212 static inline int inet_is_local_reserved_port(struct net *net, int port)
213 {
214 	if (!net->ipv4.sysctl_local_reserved_ports)
215 		return 0;
216 	return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
217 }
218 #else
219 static inline int inet_is_local_reserved_port(struct net *net, int port)
220 {
221 	return 0;
222 }
223 #endif
224 
225 extern int sysctl_ip_nonlocal_bind;
226 
227 /* From inetpeer.c */
228 extern int inet_peer_threshold;
229 extern int inet_peer_minttl;
230 extern int inet_peer_maxttl;
231 
232 /* From ip_input.c */
233 extern int sysctl_ip_early_demux;
234 
235 /* From ip_output.c */
236 extern int sysctl_ip_dynaddr;
237 
238 void ipfrag_init(void);
239 
240 void ip_static_sysctl_init(void);
241 
242 #define IP4_REPLY_MARK(net, mark) \
243 	((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
244 
245 static inline bool ip_is_fragment(const struct iphdr *iph)
246 {
247 	return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
248 }
249 
250 #ifdef CONFIG_INET
251 #include <net/dst.h>
252 
253 /* The function in 2.2 was invalid, producing wrong result for
254  * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
255 static inline
256 int ip_decrease_ttl(struct iphdr *iph)
257 {
258 	u32 check = (__force u32)iph->check;
259 	check += (__force u32)htons(0x0100);
260 	iph->check = (__force __sum16)(check + (check>=0xFFFF));
261 	return --iph->ttl;
262 }
263 
264 static inline
265 int ip_dont_fragment(struct sock *sk, struct dst_entry *dst)
266 {
267 	return  inet_sk(sk)->pmtudisc == IP_PMTUDISC_DO ||
268 		(inet_sk(sk)->pmtudisc == IP_PMTUDISC_WANT &&
269 		 !(dst_metric_locked(dst, RTAX_MTU)));
270 }
271 
272 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
273 {
274 	return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
275 	       inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
276 }
277 
278 static inline bool ip_sk_use_pmtu(const struct sock *sk)
279 {
280 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
281 }
282 
283 static inline bool ip_sk_ignore_df(const struct sock *sk)
284 {
285 	return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
286 	       inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
287 }
288 
289 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
290 						    bool forwarding)
291 {
292 	struct net *net = dev_net(dst->dev);
293 
294 	if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
295 	    dst_metric_locked(dst, RTAX_MTU) ||
296 	    !forwarding)
297 		return dst_mtu(dst);
298 
299 	return min(dst->dev->mtu, IP_MAX_MTU);
300 }
301 
302 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff *skb)
303 {
304 	if (!skb->sk || ip_sk_use_pmtu(skb->sk)) {
305 		bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
306 		return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
307 	} else {
308 		return min(skb_dst(skb)->dev->mtu, IP_MAX_MTU);
309 	}
310 }
311 
312 #define IP_IDENTS_SZ 2048u
313 extern atomic_t *ip_idents;
314 
315 static inline u32 ip_idents_reserve(u32 hash, int segs)
316 {
317 	atomic_t *id_ptr = ip_idents + hash % IP_IDENTS_SZ;
318 
319 	return atomic_add_return(segs, id_ptr) - segs;
320 }
321 
322 void __ip_select_ident(struct iphdr *iph, int segs);
323 
324 static inline void ip_select_ident_segs(struct sk_buff *skb, struct sock *sk, int segs)
325 {
326 	struct iphdr *iph = ip_hdr(skb);
327 
328 	if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
329 		/* This is only to work around buggy Windows95/2000
330 		 * VJ compression implementations.  If the ID field
331 		 * does not change, they drop every other packet in
332 		 * a TCP stream using header compression.
333 		 */
334 		if (sk && inet_sk(sk)->inet_daddr) {
335 			iph->id = htons(inet_sk(sk)->inet_id);
336 			inet_sk(sk)->inet_id += segs;
337 		} else {
338 			iph->id = 0;
339 		}
340 	} else {
341 		__ip_select_ident(iph, segs);
342 	}
343 }
344 
345 static inline void ip_select_ident(struct sk_buff *skb, struct sock *sk)
346 {
347 	ip_select_ident_segs(skb, sk, 1);
348 }
349 
350 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
351 {
352 	return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
353 				  skb->len, proto, 0);
354 }
355 
356 /*
357  *	Map a multicast IP onto multicast MAC for type ethernet.
358  */
359 
360 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
361 {
362 	__u32 addr=ntohl(naddr);
363 	buf[0]=0x01;
364 	buf[1]=0x00;
365 	buf[2]=0x5e;
366 	buf[5]=addr&0xFF;
367 	addr>>=8;
368 	buf[4]=addr&0xFF;
369 	addr>>=8;
370 	buf[3]=addr&0x7F;
371 }
372 
373 /*
374  *	Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
375  *	Leave P_Key as 0 to be filled in by driver.
376  */
377 
378 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
379 {
380 	__u32 addr;
381 	unsigned char scope = broadcast[5] & 0xF;
382 
383 	buf[0]  = 0;		/* Reserved */
384 	buf[1]  = 0xff;		/* Multicast QPN */
385 	buf[2]  = 0xff;
386 	buf[3]  = 0xff;
387 	addr    = ntohl(naddr);
388 	buf[4]  = 0xff;
389 	buf[5]  = 0x10 | scope;	/* scope from broadcast address */
390 	buf[6]  = 0x40;		/* IPv4 signature */
391 	buf[7]  = 0x1b;
392 	buf[8]  = broadcast[8];		/* P_Key */
393 	buf[9]  = broadcast[9];
394 	buf[10] = 0;
395 	buf[11] = 0;
396 	buf[12] = 0;
397 	buf[13] = 0;
398 	buf[14] = 0;
399 	buf[15] = 0;
400 	buf[19] = addr & 0xff;
401 	addr  >>= 8;
402 	buf[18] = addr & 0xff;
403 	addr  >>= 8;
404 	buf[17] = addr & 0xff;
405 	addr  >>= 8;
406 	buf[16] = addr & 0x0f;
407 }
408 
409 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
410 {
411 	if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
412 		memcpy(buf, broadcast, 4);
413 	else
414 		memcpy(buf, &naddr, sizeof(naddr));
415 }
416 
417 #if IS_ENABLED(CONFIG_IPV6)
418 #include <linux/ipv6.h>
419 #endif
420 
421 static __inline__ void inet_reset_saddr(struct sock *sk)
422 {
423 	inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
424 #if IS_ENABLED(CONFIG_IPV6)
425 	if (sk->sk_family == PF_INET6) {
426 		struct ipv6_pinfo *np = inet6_sk(sk);
427 
428 		memset(&np->saddr, 0, sizeof(np->saddr));
429 		memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
430 	}
431 #endif
432 }
433 
434 #endif
435 
436 static inline int sk_mc_loop(struct sock *sk)
437 {
438 	if (!sk)
439 		return 1;
440 	switch (sk->sk_family) {
441 	case AF_INET:
442 		return inet_sk(sk)->mc_loop;
443 #if IS_ENABLED(CONFIG_IPV6)
444 	case AF_INET6:
445 		return inet6_sk(sk)->mc_loop;
446 #endif
447 	}
448 	WARN_ON(1);
449 	return 1;
450 }
451 
452 bool ip_call_ra_chain(struct sk_buff *skb);
453 
454 /*
455  *	Functions provided by ip_fragment.c
456  */
457 
458 enum ip_defrag_users {
459 	IP_DEFRAG_LOCAL_DELIVER,
460 	IP_DEFRAG_CALL_RA_CHAIN,
461 	IP_DEFRAG_CONNTRACK_IN,
462 	__IP_DEFRAG_CONNTRACK_IN_END	= IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
463 	IP_DEFRAG_CONNTRACK_OUT,
464 	__IP_DEFRAG_CONNTRACK_OUT_END	= IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
465 	IP_DEFRAG_CONNTRACK_BRIDGE_IN,
466 	__IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
467 	IP_DEFRAG_VS_IN,
468 	IP_DEFRAG_VS_OUT,
469 	IP_DEFRAG_VS_FWD,
470 	IP_DEFRAG_AF_PACKET,
471 	IP_DEFRAG_MACVLAN,
472 };
473 
474 int ip_defrag(struct sk_buff *skb, u32 user);
475 #ifdef CONFIG_INET
476 struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user);
477 #else
478 static inline struct sk_buff *ip_check_defrag(struct sk_buff *skb, u32 user)
479 {
480 	return skb;
481 }
482 #endif
483 int ip_frag_mem(struct net *net);
484 int ip_frag_nqueues(struct net *net);
485 
486 /*
487  *	Functions provided by ip_forward.c
488  */
489 
490 int ip_forward(struct sk_buff *skb);
491 
492 /*
493  *	Functions provided by ip_options.c
494  */
495 
496 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
497 		      __be32 daddr, struct rtable *rt, int is_frag);
498 int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb);
499 void ip_options_fragment(struct sk_buff *skb);
500 int ip_options_compile(struct net *net, struct ip_options *opt,
501 		       struct sk_buff *skb);
502 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
503 		   unsigned char *data, int optlen);
504 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
505 			     unsigned char __user *data, int optlen);
506 void ip_options_undo(struct ip_options *opt);
507 void ip_forward_options(struct sk_buff *skb);
508 int ip_options_rcv_srr(struct sk_buff *skb);
509 
510 /*
511  *	Functions provided by ip_sockglue.c
512  */
513 
514 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
515 void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb);
516 int ip_cmsg_send(struct net *net, struct msghdr *msg,
517 		 struct ipcm_cookie *ipc, bool allow_ipv6);
518 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
519 		  unsigned int optlen);
520 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
521 		  int __user *optlen);
522 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
523 			 char __user *optval, unsigned int optlen);
524 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
525 			 char __user *optval, int __user *optlen);
526 int ip_ra_control(struct sock *sk, unsigned char on,
527 		  void (*destructor)(struct sock *));
528 
529 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
530 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
531 		   u32 info, u8 *payload);
532 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
533 		    u32 info);
534 
535 #ifdef CONFIG_PROC_FS
536 int ip_misc_proc_init(void);
537 #endif
538 
539 #endif	/* _IP_H */
540