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 {
42 struct ip_options opt; /* Compiled IP options */
45 #define IPSKB_FORWARDED BIT(0)
46 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
47 #define IPSKB_XFRM_TRANSFORMED BIT(2)
48 #define IPSKB_FRAG_COMPLETE BIT(3)
49 #define IPSKB_REROUTED BIT(4)
50 #define IPSKB_DOREDIRECT BIT(5)
51 #define IPSKB_FRAG_PMTU BIT(6)
52 #define IPSKB_L3SLAVE BIT(7)
57 static inline bool ipv4_l3mdev_skb(u16 flags)
59 return !!(flags & IPSKB_L3SLAVE);
62 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
64 return ip_hdr(skb)->ihl * 4;
68 struct sockcm_cookie sockc;
71 struct ip_options_rcu *opt;
78 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
79 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
82 struct ip_ra_chain __rcu *next;
85 void (*destructor)(struct sock *);
86 struct sock *saved_sk;
91 extern struct ip_ra_chain __rcu *ip_ra_chain;
94 #define IP_CE 0x8000 /* Flag: "Congestion" */
95 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
96 #define IP_MF 0x2000 /* Flag: "More Fragments" */
97 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
99 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
107 int igmp_mc_init(void);
110 * Functions provided by ip.c
113 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
114 __be32 saddr, __be32 daddr,
115 struct ip_options_rcu *opt);
116 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
117 struct net_device *orig_dev);
118 int ip_local_deliver(struct sk_buff *skb);
119 int ip_mr_input(struct sk_buff *skb);
120 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
121 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
122 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
123 int (*output)(struct net *, struct sock *, struct sk_buff *));
124 void ip_send_check(struct iphdr *ip);
125 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
126 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
128 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
130 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
131 int getfrag(void *from, char *to, int offset, int len,
132 int odd, struct sk_buff *skb),
133 void *from, int len, int protolen,
134 struct ipcm_cookie *ipc,
137 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
138 struct sk_buff *skb);
139 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
140 int offset, size_t size, int flags);
141 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
142 struct sk_buff_head *queue,
143 struct inet_cork *cork);
144 int ip_send_skb(struct net *net, struct sk_buff *skb);
145 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
146 void ip_flush_pending_frames(struct sock *sk);
147 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
148 int getfrag(void *from, char *to, int offset,
149 int len, int odd, struct sk_buff *skb),
150 void *from, int length, int transhdrlen,
151 struct ipcm_cookie *ipc, struct rtable **rtp,
154 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
156 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
159 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
161 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
164 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
166 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
170 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
171 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
173 void ip4_datagram_release_cb(struct sock *sk);
175 struct ip_reply_arg {
179 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
180 /* -1 if not needed */
185 #define IP_REPLY_ARG_NOSRCCHECK 1
187 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
189 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
192 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
193 const struct ip_options *sopt,
194 __be32 daddr, __be32 saddr,
195 const struct ip_reply_arg *arg,
198 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
199 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
200 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
201 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
202 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
203 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
204 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
205 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
206 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
207 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
209 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
210 unsigned long snmp_fold_field(void __percpu *mib, int offt);
211 #if BITS_PER_LONG==32
212 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
213 size_t syncp_offset);
214 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
216 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
219 return snmp_get_cpu_field(mib, cpu, offct);
223 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
225 return snmp_fold_field(mib, offt);
229 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
232 for_each_possible_cpu(c) { \
233 for (i = 0; stats_list[i].name; i++) \
234 buff64[i] += snmp_get_cpu_field64( \
236 c, stats_list[i].entry, \
241 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
244 for_each_possible_cpu(c) { \
245 for (i = 0; stats_list[i].name; i++) \
246 buff[i] += snmp_get_cpu_field( \
248 c, stats_list[i].entry); \
252 void inet_get_local_port_range(struct net *net, int *low, int *high);
255 static inline int inet_is_local_reserved_port(struct net *net, int port)
257 if (!net->ipv4.sysctl_local_reserved_ports)
259 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
262 static inline bool sysctl_dev_name_is_allowed(const char *name)
264 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
268 static inline int inet_is_local_reserved_port(struct net *net, int port)
274 __be32 inet_current_timestamp(void);
276 /* From inetpeer.c */
277 extern int inet_peer_threshold;
278 extern int inet_peer_minttl;
279 extern int inet_peer_maxttl;
281 void ipfrag_init(void);
283 void ip_static_sysctl_init(void);
285 #define IP4_REPLY_MARK(net, mark) \
286 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
288 static inline bool ip_is_fragment(const struct iphdr *iph)
290 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
296 /* The function in 2.2 was invalid, producing wrong result for
297 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
299 int ip_decrease_ttl(struct iphdr *iph)
301 u32 check = (__force u32)iph->check;
302 check += (__force u32)htons(0x0100);
303 iph->check = (__force __sum16)(check + (check>=0xFFFF));
307 static inline int ip_mtu_locked(const struct dst_entry *dst)
309 const struct rtable *rt = (const struct rtable *)dst;
311 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
315 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
317 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
319 return pmtudisc == IP_PMTUDISC_DO ||
320 (pmtudisc == IP_PMTUDISC_WANT &&
321 !ip_mtu_locked(dst));
324 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
326 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
327 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
330 static inline bool ip_sk_use_pmtu(const struct sock *sk)
332 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
335 static inline bool ip_sk_ignore_df(const struct sock *sk)
337 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
338 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
341 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
344 struct net *net = dev_net(dst->dev);
347 if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
348 ip_mtu_locked(dst) ||
352 /* 'forwarding = true' case should always honour route mtu */
353 mtu = dst_metric_raw(dst, RTAX_MTU);
357 return min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU);
360 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
361 const struct sk_buff *skb)
363 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
364 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
366 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
369 return min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
372 u32 ip_idents_reserve(u32 hash, int segs);
373 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
375 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
376 struct sock *sk, int segs)
378 struct iphdr *iph = ip_hdr(skb);
380 /* We had many attacks based on IPID, use the private
381 * generator as much as we can.
383 if (sk && inet_sk(sk)->inet_daddr) {
384 iph->id = htons(inet_sk(sk)->inet_id);
385 inet_sk(sk)->inet_id += segs;
388 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
391 /* Unfortunately we need the big hammer to get a suitable IPID */
392 __ip_select_ident(net, iph, segs);
396 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
399 ip_select_ident_segs(net, skb, sk, 1);
402 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
404 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
408 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
409 * Equivalent to : flow->v4addrs.src = iph->saddr;
410 * flow->v4addrs.dst = iph->daddr;
412 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
413 const struct iphdr *iph)
415 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
416 offsetof(typeof(flow->addrs), v4addrs.src) +
417 sizeof(flow->addrs.v4addrs.src));
418 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs));
419 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
422 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
424 const struct iphdr *iph = skb_gro_network_header(skb);
426 return csum_tcpudp_nofold(iph->saddr, iph->daddr,
427 skb_gro_len(skb), proto, 0);
431 * Map a multicast IP onto multicast MAC for type ethernet.
434 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
436 __u32 addr=ntohl(naddr);
448 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
449 * Leave P_Key as 0 to be filled in by driver.
452 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
455 unsigned char scope = broadcast[5] & 0xF;
457 buf[0] = 0; /* Reserved */
458 buf[1] = 0xff; /* Multicast QPN */
463 buf[5] = 0x10 | scope; /* scope from broadcast address */
464 buf[6] = 0x40; /* IPv4 signature */
466 buf[8] = broadcast[8]; /* P_Key */
467 buf[9] = broadcast[9];
474 buf[19] = addr & 0xff;
476 buf[18] = addr & 0xff;
478 buf[17] = addr & 0xff;
480 buf[16] = addr & 0x0f;
483 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
485 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
486 memcpy(buf, broadcast, 4);
488 memcpy(buf, &naddr, sizeof(naddr));
491 #if IS_ENABLED(CONFIG_IPV6)
492 #include <linux/ipv6.h>
495 static __inline__ void inet_reset_saddr(struct sock *sk)
497 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
498 #if IS_ENABLED(CONFIG_IPV6)
499 if (sk->sk_family == PF_INET6) {
500 struct ipv6_pinfo *np = inet6_sk(sk);
502 memset(&np->saddr, 0, sizeof(np->saddr));
503 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
510 static inline unsigned int ipv4_addr_hash(__be32 ip)
512 return (__force unsigned int) ip;
515 bool ip_call_ra_chain(struct sk_buff *skb);
518 * Functions provided by ip_fragment.c
521 enum ip_defrag_users {
522 IP_DEFRAG_LOCAL_DELIVER,
523 IP_DEFRAG_CALL_RA_CHAIN,
524 IP_DEFRAG_CONNTRACK_IN,
525 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
526 IP_DEFRAG_CONNTRACK_OUT,
527 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
528 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
529 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
537 /* Return true if the value of 'user' is between 'lower_bond'
538 * and 'upper_bond' inclusively.
540 static inline bool ip_defrag_user_in_between(u32 user,
541 enum ip_defrag_users lower_bond,
542 enum ip_defrag_users upper_bond)
544 return user >= lower_bond && user <= upper_bond;
547 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
549 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
551 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
558 * Functions provided by ip_forward.c
561 int ip_forward(struct sk_buff *skb);
564 * Functions provided by ip_options.c
567 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
568 __be32 daddr, struct rtable *rt, int is_frag);
570 int __ip_options_echo(struct ip_options *dopt, struct sk_buff *skb,
571 const struct ip_options *sopt);
572 static inline int ip_options_echo(struct ip_options *dopt, struct sk_buff *skb)
574 return __ip_options_echo(dopt, skb, &IPCB(skb)->opt);
577 void ip_options_fragment(struct sk_buff *skb);
578 int __ip_options_compile(struct net *net, struct ip_options *opt,
579 struct sk_buff *skb, __be32 *info);
580 int ip_options_compile(struct net *net, struct ip_options *opt,
581 struct sk_buff *skb);
582 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
583 unsigned char *data, int optlen);
584 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
585 unsigned char __user *data, int optlen);
586 void ip_options_undo(struct ip_options *opt);
587 void ip_forward_options(struct sk_buff *skb);
588 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
591 * Functions provided by ip_sockglue.c
594 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
595 void ip_cmsg_recv_offset(struct msghdr *msg, struct sk_buff *skb, int tlen, int offset);
596 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
597 struct ipcm_cookie *ipc, bool allow_ipv6);
598 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
599 unsigned int optlen);
600 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
602 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
603 char __user *optval, unsigned int optlen);
604 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
605 char __user *optval, int __user *optlen);
606 int ip_ra_control(struct sock *sk, unsigned char on,
607 void (*destructor)(struct sock *));
609 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
610 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
611 u32 info, u8 *payload);
612 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
615 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
617 ip_cmsg_recv_offset(msg, skb, 0, 0);
620 bool icmp_global_allow(void);
621 extern int sysctl_icmp_msgs_per_sec;
622 extern int sysctl_icmp_msgs_burst;
624 #ifdef CONFIG_PROC_FS
625 int ip_misc_proc_init(void);
628 static inline bool inetdev_valid_mtu(unsigned int mtu)
630 return likely(mtu >= IPV4_MIN_MTU);