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
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>
29 #include <linux/jhash.h>
31 #include <net/inet_sock.h>
32 #include <net/route.h>
35 #include <net/flow_dissector.h>
36 #include <net/netns/hash.h>
38 #define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */
39 #define IPV4_MIN_MTU 68 /* RFC 791 */
43 struct inet_skb_parm {
45 struct ip_options opt; /* Compiled IP options */
48 #define IPSKB_FORWARDED BIT(0)
49 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
50 #define IPSKB_XFRM_TRANSFORMED BIT(2)
51 #define IPSKB_FRAG_COMPLETE BIT(3)
52 #define IPSKB_REROUTED BIT(4)
53 #define IPSKB_DOREDIRECT BIT(5)
54 #define IPSKB_FRAG_PMTU BIT(6)
55 #define IPSKB_L3SLAVE BIT(7)
60 static inline bool ipv4_l3mdev_skb(u16 flags)
62 return !!(flags & IPSKB_L3SLAVE);
65 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
67 return ip_hdr(skb)->ihl * 4;
71 struct sockcm_cookie sockc;
74 struct ip_options_rcu *opt;
82 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
83 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
85 /* return enslaved device index if relevant */
86 static inline int inet_sdif(struct sk_buff *skb)
88 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
89 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
90 return IPCB(skb)->iif;
95 /* Special input handler for packets caught by router alert option.
96 They are selected only by protocol field, and then processed likely
97 local ones; but only if someone wants them! Otherwise, router
98 not running rsvpd will kill RSVP.
100 It is user level problem, what it will make with them.
101 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
102 but receiver should be enough clever f.e. to forward mtrace requests,
103 sent to multicast group to reach destination designated router.
107 struct ip_ra_chain __rcu *next;
110 void (*destructor)(struct sock *);
111 struct sock *saved_sk;
117 #define IP_CE 0x8000 /* Flag: "Congestion" */
118 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
119 #define IP_MF 0x2000 /* Flag: "More Fragments" */
120 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
122 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
130 int igmp_mc_init(void);
133 * Functions provided by ip.c
136 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
137 __be32 saddr, __be32 daddr,
138 struct ip_options_rcu *opt);
139 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
140 struct net_device *orig_dev);
141 int ip_local_deliver(struct sk_buff *skb);
142 int ip_mr_input(struct sk_buff *skb);
143 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
144 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
145 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
146 int (*output)(struct net *, struct sock *, struct sk_buff *));
147 void ip_send_check(struct iphdr *ip);
148 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
149 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
151 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
153 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
154 int getfrag(void *from, char *to, int offset, int len,
155 int odd, struct sk_buff *skb),
156 void *from, int len, int protolen,
157 struct ipcm_cookie *ipc,
160 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
161 struct sk_buff *skb);
162 ssize_t ip_append_page(struct sock *sk, struct flowi4 *fl4, struct page *page,
163 int offset, size_t size, int flags);
164 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
165 struct sk_buff_head *queue,
166 struct inet_cork *cork);
167 int ip_send_skb(struct net *net, struct sk_buff *skb);
168 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
169 void ip_flush_pending_frames(struct sock *sk);
170 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
171 int getfrag(void *from, char *to, int offset,
172 int len, int odd, struct sk_buff *skb),
173 void *from, int length, int transhdrlen,
174 struct ipcm_cookie *ipc, struct rtable **rtp,
175 struct inet_cork *cork, unsigned int flags);
177 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
179 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
182 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
184 return (ipc->tos != -1) ? RT_TOS(ipc->tos) : RT_TOS(inet->tos);
187 static inline __u8 get_rtconn_flags(struct ipcm_cookie* ipc, struct sock* sk)
189 return (ipc->tos != -1) ? RT_CONN_FLAGS_TOS(sk, ipc->tos) : RT_CONN_FLAGS(sk);
193 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
194 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
196 void ip4_datagram_release_cb(struct sock *sk);
198 struct ip_reply_arg {
202 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
203 /* -1 if not needed */
209 #define IP_REPLY_ARG_NOSRCCHECK 1
211 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
213 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
216 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
217 const struct ip_options *sopt,
218 __be32 daddr, __be32 saddr,
219 const struct ip_reply_arg *arg,
222 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
223 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
224 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
225 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
226 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
227 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
228 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
229 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
230 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
231 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
233 u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offct);
234 unsigned long snmp_fold_field(void __percpu *mib, int offt);
235 #if BITS_PER_LONG==32
236 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
237 size_t syncp_offset);
238 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
240 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
243 return snmp_get_cpu_field(mib, cpu, offct);
247 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
249 return snmp_fold_field(mib, offt);
253 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
256 for_each_possible_cpu(c) { \
257 for (i = 0; stats_list[i].name; i++) \
258 buff64[i] += snmp_get_cpu_field64( \
260 c, stats_list[i].entry, \
265 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
268 for_each_possible_cpu(c) { \
269 for (i = 0; stats_list[i].name; i++) \
270 buff[i] += snmp_get_cpu_field( \
272 c, stats_list[i].entry); \
276 void inet_get_local_port_range(struct net *net, int *low, int *high);
279 static inline int inet_is_local_reserved_port(struct net *net, int port)
281 if (!net->ipv4.sysctl_local_reserved_ports)
283 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
286 static inline bool sysctl_dev_name_is_allowed(const char *name)
288 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
291 static inline int inet_prot_sock(struct net *net)
293 return net->ipv4.sysctl_ip_prot_sock;
297 static inline int inet_is_local_reserved_port(struct net *net, int port)
302 static inline int inet_prot_sock(struct net *net)
308 __be32 inet_current_timestamp(void);
310 /* From inetpeer.c */
311 extern int inet_peer_threshold;
312 extern int inet_peer_minttl;
313 extern int inet_peer_maxttl;
315 void ipfrag_init(void);
317 void ip_static_sysctl_init(void);
319 #define IP4_REPLY_MARK(net, mark) \
320 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
322 static inline bool ip_is_fragment(const struct iphdr *iph)
324 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
330 /* The function in 2.2 was invalid, producing wrong result for
331 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
333 int ip_decrease_ttl(struct iphdr *iph)
335 u32 check = (__force u32)iph->check;
336 check += (__force u32)htons(0x0100);
337 iph->check = (__force __sum16)(check + (check>=0xFFFF));
341 static inline int ip_mtu_locked(const struct dst_entry *dst)
343 const struct rtable *rt = (const struct rtable *)dst;
345 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
349 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
351 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
353 return pmtudisc == IP_PMTUDISC_DO ||
354 (pmtudisc == IP_PMTUDISC_WANT &&
355 !ip_mtu_locked(dst));
358 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
360 return inet_sk(sk)->pmtudisc != IP_PMTUDISC_INTERFACE &&
361 inet_sk(sk)->pmtudisc != IP_PMTUDISC_OMIT;
364 static inline bool ip_sk_use_pmtu(const struct sock *sk)
366 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_PROBE;
369 static inline bool ip_sk_ignore_df(const struct sock *sk)
371 return inet_sk(sk)->pmtudisc < IP_PMTUDISC_DO ||
372 inet_sk(sk)->pmtudisc == IP_PMTUDISC_OMIT;
375 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
378 struct net *net = dev_net(dst->dev);
380 if (net->ipv4.sysctl_ip_fwd_use_pmtu ||
381 ip_mtu_locked(dst) ||
385 return min(READ_ONCE(dst->dev->mtu), IP_MAX_MTU);
388 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
389 const struct sk_buff *skb)
391 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
392 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
394 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
397 return min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
400 int ip_metrics_convert(struct net *net, struct nlattr *fc_mx, int fc_mx_len,
403 u32 ip_idents_reserve(u32 hash, int segs);
404 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
406 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
407 struct sock *sk, int segs)
409 struct iphdr *iph = ip_hdr(skb);
411 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
412 /* This is only to work around buggy Windows95/2000
413 * VJ compression implementations. If the ID field
414 * does not change, they drop every other packet in
415 * a TCP stream using header compression.
417 if (sk && inet_sk(sk)->inet_daddr) {
418 iph->id = htons(inet_sk(sk)->inet_id);
419 inet_sk(sk)->inet_id += segs;
424 __ip_select_ident(net, iph, segs);
428 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
431 ip_select_ident_segs(net, skb, sk, 1);
434 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
436 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
440 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
441 * Equivalent to : flow->v4addrs.src = iph->saddr;
442 * flow->v4addrs.dst = iph->daddr;
444 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
445 const struct iphdr *iph)
447 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
448 offsetof(typeof(flow->addrs), v4addrs.src) +
449 sizeof(flow->addrs.v4addrs.src));
450 memcpy(&flow->addrs.v4addrs, &iph->saddr, sizeof(flow->addrs.v4addrs));
451 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
454 static inline __wsum inet_gro_compute_pseudo(struct sk_buff *skb, int proto)
456 const struct iphdr *iph = skb_gro_network_header(skb);
458 return csum_tcpudp_nofold(iph->saddr, iph->daddr,
459 skb_gro_len(skb), proto, 0);
463 * Map a multicast IP onto multicast MAC for type ethernet.
466 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
468 __u32 addr=ntohl(naddr);
480 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
481 * Leave P_Key as 0 to be filled in by driver.
484 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
487 unsigned char scope = broadcast[5] & 0xF;
489 buf[0] = 0; /* Reserved */
490 buf[1] = 0xff; /* Multicast QPN */
495 buf[5] = 0x10 | scope; /* scope from broadcast address */
496 buf[6] = 0x40; /* IPv4 signature */
498 buf[8] = broadcast[8]; /* P_Key */
499 buf[9] = broadcast[9];
506 buf[19] = addr & 0xff;
508 buf[18] = addr & 0xff;
510 buf[17] = addr & 0xff;
512 buf[16] = addr & 0x0f;
515 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
517 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
518 memcpy(buf, broadcast, 4);
520 memcpy(buf, &naddr, sizeof(naddr));
523 #if IS_ENABLED(CONFIG_IPV6)
524 #include <linux/ipv6.h>
527 static __inline__ void inet_reset_saddr(struct sock *sk)
529 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
530 #if IS_ENABLED(CONFIG_IPV6)
531 if (sk->sk_family == PF_INET6) {
532 struct ipv6_pinfo *np = inet6_sk(sk);
534 memset(&np->saddr, 0, sizeof(np->saddr));
535 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
542 static inline unsigned int ipv4_addr_hash(__be32 ip)
544 return (__force unsigned int) ip;
547 static inline u32 ipv4_portaddr_hash(const struct net *net,
551 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
554 bool ip_call_ra_chain(struct sk_buff *skb);
557 * Functions provided by ip_fragment.c
560 enum ip_defrag_users {
561 IP_DEFRAG_LOCAL_DELIVER,
562 IP_DEFRAG_CALL_RA_CHAIN,
563 IP_DEFRAG_CONNTRACK_IN,
564 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
565 IP_DEFRAG_CONNTRACK_OUT,
566 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
567 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
568 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
576 /* Return true if the value of 'user' is between 'lower_bond'
577 * and 'upper_bond' inclusively.
579 static inline bool ip_defrag_user_in_between(u32 user,
580 enum ip_defrag_users lower_bond,
581 enum ip_defrag_users upper_bond)
583 return user >= lower_bond && user <= upper_bond;
586 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
588 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
590 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
597 * Functions provided by ip_forward.c
600 int ip_forward(struct sk_buff *skb);
603 * Functions provided by ip_options.c
606 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
607 __be32 daddr, struct rtable *rt, int is_frag);
609 int __ip_options_echo(struct net *net, struct ip_options *dopt,
610 struct sk_buff *skb, const struct ip_options *sopt);
611 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
614 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
617 void ip_options_fragment(struct sk_buff *skb);
618 int ip_options_compile(struct net *net, struct ip_options *opt,
619 struct sk_buff *skb);
620 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
621 unsigned char *data, int optlen);
622 int ip_options_get_from_user(struct net *net, struct ip_options_rcu **optp,
623 unsigned char __user *data, int optlen);
624 void ip_options_undo(struct ip_options *opt);
625 void ip_forward_options(struct sk_buff *skb);
626 int ip_options_rcv_srr(struct sk_buff *skb);
629 * Functions provided by ip_sockglue.c
632 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb);
633 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
634 struct sk_buff *skb, int tlen, int offset);
635 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
636 struct ipcm_cookie *ipc, bool allow_ipv6);
637 int ip_setsockopt(struct sock *sk, int level, int optname, char __user *optval,
638 unsigned int optlen);
639 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
641 int compat_ip_setsockopt(struct sock *sk, int level, int optname,
642 char __user *optval, unsigned int optlen);
643 int compat_ip_getsockopt(struct sock *sk, int level, int optname,
644 char __user *optval, int __user *optlen);
645 int ip_ra_control(struct sock *sk, unsigned char on,
646 void (*destructor)(struct sock *));
648 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
649 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
650 u32 info, u8 *payload);
651 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
654 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
656 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
659 bool icmp_global_allow(void);
660 extern int sysctl_icmp_msgs_per_sec;
661 extern int sysctl_icmp_msgs_burst;
663 #ifdef CONFIG_PROC_FS
664 int ip_misc_proc_init(void);
667 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto,
668 struct netlink_ext_ack *extack);