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.
6 * Definitions for the IP module.
8 * Version: @(#)ip.h 1.0.2 05/07/93
11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * Alan Cox, <gw4pts@gw4pts.ampr.org>
15 * Mike McLagan : Routing by source
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.
25 #include <linux/types.h>
28 #include <linux/skbuff.h>
30 #include <net/inet_sock.h>
31 #include <net/route.h>
34 #include <net/flow_dissector.h>
36 #define IPV4_MIN_MTU 68 /* RFC 791 */
40 struct inet_skb_parm {
41 struct ip_options opt; /* Compiled IP options */
44 #define IPSKB_FORWARDED BIT(0)
45 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
46 #define IPSKB_XFRM_TRANSFORMED BIT(2)
47 #define IPSKB_FRAG_COMPLETE BIT(3)
48 #define IPSKB_REROUTED BIT(4)
49 #define IPSKB_DOREDIRECT BIT(5)
50 #define IPSKB_FRAG_PMTU BIT(6)
55 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
57 return ip_hdr(skb)->ihl * 4;
63 struct ip_options_rcu *opt;
70 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
71 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
74 struct ip_ra_chain __rcu *next;
77 void (*destructor)(struct sock *);
78 struct sock *saved_sk;
83 extern struct ip_ra_chain __rcu *ip_ra_chain;
86 #define IP_CE 0x8000 /* Flag: "Congestion" */
87 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
88 #define IP_MF 0x2000 /* Flag: "More Fragments" */
89 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
91 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
99 int igmp_mc_init(void);
102 * Functions provided by ip.c
105 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
106 __be32 saddr, __be32 daddr,
107 struct ip_options_rcu *opt);
108 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
109 struct net_device *orig_dev);
110 int ip_local_deliver(struct sk_buff *skb);
111 int ip_mr_input(struct sk_buff *skb);
112 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
113 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
114 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
115 int (*output)(struct net *, struct sock *, struct sk_buff *));
116 void ip_send_check(struct iphdr *ip);
117 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
118 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
120 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
122 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
123 int getfrag(void *from, char *to, int offset, int len,
124 int odd, struct sk_buff *skb),
125 void *from, int len, int protolen,
126 struct ipcm_cookie *ipc,
129 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
130 struct sk_buff *skb);
131 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
132 int offset, size_t size, int flags);
133 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
134 struct sk_buff_head *queue,
135 struct inet_cork *cork);
136 int ip_send_skb(struct net *net, struct sk_buff *skb);
137 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
138 void ip_flush_pending_frames(struct sock *sk);
139 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
140 int getfrag(void *from, char *to, int offset,
141 int len, int odd, struct sk_buff *skb),
142 void *from, int length, int transhdrlen,
143 struct ipcm_cookie *ipc, struct rtable **rtp,
146 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
148 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
151 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
153 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
156 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
158 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
162 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
163 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
165 void ip4_datagram_release_cb(struct sock *sk);
167 struct ip_reply_arg {
171 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
172 /* -1 if not needed */
177 #define IP_REPLY_ARG_NOSRCCHECK 1
179 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
181 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
184 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
185 const struct ip_options *sopt,
186 __be32 daddr, __be32 saddr,
187 const struct ip_reply_arg *arg,
190 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
191 #define IP_INC_STATS_BH(net, field) SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
192 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
193 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
194 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
195 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
196 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
197 #define NET_INC_STATS_BH(net, field) SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
198 #define NET_INC_STATS_USER(net, field) SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
199 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
200 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
201 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
203 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
204 unsigned long snmp_fold_field(void __percpu *mib, int offt);
205 #if BITS_PER_LONG==32
206 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
207 size_t syncp_offset);
208 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
210 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
213 return snmp_get_cpu_field(mib, cpu, offct);
217 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
219 return snmp_fold_field(mib, offt);
223 void inet_get_local_port_range(struct net *net, int *low, int *high);
226 static inline int inet_is_local_reserved_port(struct net *net, int port)
228 if (!net->ipv4.sysctl_local_reserved_ports)
230 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
233 static inline bool sysctl_dev_name_is_allowed(const char *name)
235 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
239 static inline int inet_is_local_reserved_port(struct net *net, int port)
245 /* From inetpeer.c */
246 extern int inet_peer_threshold;
247 extern int inet_peer_minttl;
248 extern int inet_peer_maxttl;
250 /* From ip_input.c */
251 extern int sysctl_ip_early_demux;
253 /* From ip_output.c */
254 extern int sysctl_ip_dynaddr;
256 void ipfrag_init(void);
258 void ip_static_sysctl_init(void);
260 #define IP4_REPLY_MARK(net, mark) \
261 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
263 static inline bool ip_is_fragment(const struct iphdr *iph)
265 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
271 /* The function in 2.2 was invalid, producing wrong result for
272 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
274 int ip_decrease_ttl(struct iphdr *iph)
276 u32 check = (__force u32)iph->check;
277 check += (__force u32)htons(0x0100);
278 iph->check = (__force __sum16)(check + (check>=0xFFFF));
282 static inline int ip_mtu_locked(const struct dst_entry *dst)
284 const struct rtable *rt = (const struct rtable *)dst;
286 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
290 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
292 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
294 return pmtudisc == IP_PMTUDISC_DO ||
295 (pmtudisc == IP_PMTUDISC_WANT &&
296 !ip_mtu_locked(dst));
299 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
301 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
302 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
305 static inline bool ip_sk_use_pmtu(const struct sock *sk)
307 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
310 static inline bool ip_sk_ignore_df(const struct sock *sk)
312 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
313 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
316 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
319 struct net *net = dev_net(dst->dev);
322 if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
323 ip_mtu_locked(dst) ||
327 /* 'forwarding = true' case should always honour route mtu */
328 mtu = dst_metric_raw(dst, RTAX_MTU);
332 return min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU);
335 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff *skb)
337 struct sock *sk = skb->sk;
339 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
340 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
342 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
345 return min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
348 u32 ip_idents_reserve(u32 hash, int segs);
349 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
351 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
352 struct sock *sk, int segs)
354 struct iphdr *iph = ip_hdr(skb);
356 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
357 /* This is only to work around buggy Windows95/2000
358 * VJ compression implementations. If the ID field
359 * does not change, they drop every other packet in
360 * a TCP stream using header compression.
362 if (sk && inet_sk(sk)->inet_daddr) {
363 iph->id = htons(inet_sk(sk)->inet_id);
364 inet_sk(sk)->inet_id += segs;
369 __ip_select_ident(net, iph, segs);
373 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
376 ip_select_ident_segs(net, skb, sk, 1);
379 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
381 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
385 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
386 * Equivalent to : flow->v4addrs.src = iph->saddr;
387 * flow->v4addrs.dst = iph->daddr;
389 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
390 const struct iphdr *iph)
392 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
393 offsetof(typeof(flow->addrs), v4addrs.src) +
394 sizeof(flow->addrs.v4addrs.src));
395 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs));
396 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
399 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
401 const struct iphdr *iph = skb_gro_network_header(skb);
403 return csum_tcpudp_nofold(iph->saddr, iph->daddr,
404 skb_gro_len(skb), proto, 0);
408 * Map a multicast IP onto multicast MAC for type ethernet.
411 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
413 __u32 addr=ntohl(naddr);
425 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
426 * Leave P_Key as 0 to be filled in by driver.
429 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
432 unsigned char scope = broadcast[5] & 0xF;
434 buf[0] = 0; /* Reserved */
435 buf[1] = 0xff; /* Multicast QPN */
440 buf[5] = 0x10 | scope; /* scope from broadcast address */
441 buf[6] = 0x40; /* IPv4 signature */
443 buf[8] = broadcast[8]; /* P_Key */
444 buf[9] = broadcast[9];
451 buf[19] = addr & 0xff;
453 buf[18] = addr & 0xff;
455 buf[17] = addr & 0xff;
457 buf[16] = addr & 0x0f;
460 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
462 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
463 memcpy(buf, broadcast, 4);
465 memcpy(buf, &naddr, sizeof(naddr));
468 #if IS_ENABLED(CONFIG_IPV6)
469 #include <linux/ipv6.h>
472 static __inline__ void inet_reset_saddr(struct sock *sk)
474 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
475 #if IS_ENABLED(CONFIG_IPV6)
476 if (sk->sk_family == PF_INET6) {
477 struct ipv6_pinfo *np = inet6_sk(sk);
479 memset(&np->saddr, 0, sizeof(np->saddr));
480 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
487 static inline unsigned int ipv4_addr_hash(__be32 ip)
489 return (__force unsigned int) ip;
492 bool ip_call_ra_chain(struct sk_buff *skb);
495 * Functions provided by ip_fragment.c
498 enum ip_defrag_users {
499 IP_DEFRAG_LOCAL_DELIVER,
500 IP_DEFRAG_CALL_RA_CHAIN,
501 IP_DEFRAG_CONNTRACK_IN,
502 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
503 IP_DEFRAG_CONNTRACK_OUT,
504 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
505 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
506 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
514 /* Return true if the value of 'user' is between 'lower_bond'
515 * and 'upper_bond' inclusively.
517 static inline bool ip_defrag_user_in_between(u32 user,
518 enum ip_defrag_users lower_bond,
519 enum ip_defrag_users upper_bond)
521 return user >= lower_bond && user <= upper_bond;
524 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
526 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
528 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
535 * Functions provided by ip_forward.c
538 int ip_forward(struct sk_buff *skb);
541 * Functions provided by ip_options.c
544 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
545 __be32 daddr, struct rtable *rt, int is_frag);
547 int __ip_options_echo(struct ip_options *dopt, struct sk_buff *skb,
548 const struct ip_options *sopt);
549 static inline int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb)
551 return __ip_options_echo(dopt, skb, &IPCB(skb)->opt);
554 void ip_options_fragment(struct sk_buff *skb);
555 int __ip_options_compile(struct net *net, struct ip_options *opt,
556 struct sk_buff *skb, __be32 *info);
557 int ip_options_compile(struct net *net, struct ip_options *opt,
558 struct sk_buff *skb);
559 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
560 unsigned char *data, int optlen);
561 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
562 unsigned char __user *data, int optlen);
563 void ip_options_undo(struct ip_options *opt);
564 void ip_forward_options(struct sk_buff *skb);
565 int ip_options_rcv_srr(struct sk_buff *skb);
568 * Functions provided by ip_sockglue.c
571 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
572 void ip_cmsg_recv_offset(struct msghdr *msg, struct sk_buff *skb, int tlen, int offset);
573 int ip_cmsg_send(struct net *net, struct msghdr *msg,
574 struct ipcm_cookie *ipc, bool allow_ipv6);
575 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
576 unsigned int optlen);
577 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
579 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
580 char __user *optval, unsigned int optlen);
581 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
582 char __user *optval, int __user *optlen);
583 int ip_ra_control(struct sock *sk, unsigned char on,
584 void (*destructor)(struct sock *));
586 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
587 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
588 u32 info, u8 *payload);
589 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
592 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
594 ip_cmsg_recv_offset(msg, skb, 0, 0);
597 bool icmp_global_allow(void);
598 extern int sysctl_icmp_msgs_per_sec;
599 extern int sysctl_icmp_msgs_burst;
601 #ifdef CONFIG_PROC_FS
602 int ip_misc_proc_init(void);
605 static inline bool inetdev_valid_mtu(unsigned int mtu)
607 return likely(mtu >= IPV4_MIN_MTU);