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>
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
;
81 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
82 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
84 /* return enslaved device index if relevant */
85 static inline int inet_sdif(struct sk_buff
*skb
)
87 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
88 if (skb
&& ipv4_l3mdev_skb(IPCB(skb
)->flags
))
89 return IPCB(skb
)->iif
;
95 struct ip_ra_chain __rcu
*next
;
98 void (*destructor
)(struct sock
*);
99 struct sock
*saved_sk
;
104 extern struct ip_ra_chain __rcu
*ip_ra_chain
;
107 #define IP_CE 0x8000 /* Flag: "Congestion" */
108 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
109 #define IP_MF 0x2000 /* Flag: "More Fragments" */
110 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
112 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
120 int igmp_mc_init(void);
123 * Functions provided by ip.c
126 int ip_build_and_send_pkt(struct sk_buff
*skb
, const struct sock
*sk
,
127 __be32 saddr
, __be32 daddr
,
128 struct ip_options_rcu
*opt
);
129 int ip_rcv(struct sk_buff
*skb
, struct net_device
*dev
, struct packet_type
*pt
,
130 struct net_device
*orig_dev
);
131 int ip_local_deliver(struct sk_buff
*skb
);
132 int ip_mr_input(struct sk_buff
*skb
);
133 int ip_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
134 int ip_mc_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
135 int ip_do_fragment(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
,
136 int (*output
)(struct net
*, struct sock
*, struct sk_buff
*));
137 void ip_send_check(struct iphdr
*ip
);
138 int __ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
139 int ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
141 int ip_queue_xmit(struct sock
*sk
, struct sk_buff
*skb
, struct flowi
*fl
);
143 int ip_append_data(struct sock
*sk
, struct flowi4
*fl4
,
144 int getfrag(void *from
, char *to
, int offset
, int len
,
145 int odd
, struct sk_buff
*skb
),
146 void *from
, int len
, int protolen
,
147 struct ipcm_cookie
*ipc
,
150 int ip_generic_getfrag(void *from
, char *to
, int offset
, int len
, int odd
,
151 struct sk_buff
*skb
);
152 ssize_t
ip_append_page(struct sock
*sk
, struct flowi4
*fl4
, struct page
*page
,
153 int offset
, size_t size
, int flags
);
154 struct sk_buff
*__ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
155 struct sk_buff_head
*queue
,
156 struct inet_cork
*cork
);
157 int ip_send_skb(struct net
*net
, struct sk_buff
*skb
);
158 int ip_push_pending_frames(struct sock
*sk
, struct flowi4
*fl4
);
159 void ip_flush_pending_frames(struct sock
*sk
);
160 struct sk_buff
*ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
161 int getfrag(void *from
, char *to
, int offset
,
162 int len
, int odd
, struct sk_buff
*skb
),
163 void *from
, int length
, int transhdrlen
,
164 struct ipcm_cookie
*ipc
, struct rtable
**rtp
,
167 static inline struct sk_buff
*ip_finish_skb(struct sock
*sk
, struct flowi4
*fl4
)
169 return __ip_make_skb(sk
, fl4
, &sk
->sk_write_queue
, &inet_sk(sk
)->cork
.base
);
172 static inline __u8
get_rttos(struct ipcm_cookie
* ipc
, struct inet_sock
*inet
)
174 return (ipc
->tos
!= -1) ? RT_TOS(ipc
->tos
) : RT_TOS(inet
->tos
);
177 static inline __u8
get_rtconn_flags(struct ipcm_cookie
* ipc
, struct sock
* sk
)
179 return (ipc
->tos
!= -1) ? RT_CONN_FLAGS_TOS(sk
, ipc
->tos
) : RT_CONN_FLAGS(sk
);
183 int __ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
184 int ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
186 void ip4_datagram_release_cb(struct sock
*sk
);
188 struct ip_reply_arg
{
192 int csumoffset
; /* u16 offset of csum in iov[0].iov_base */
193 /* -1 if not needed */
199 #define IP_REPLY_ARG_NOSRCCHECK 1
201 static inline __u8
ip_reply_arg_flowi_flags(const struct ip_reply_arg
*arg
)
203 return (arg
->flags
& IP_REPLY_ARG_NOSRCCHECK
) ? FLOWI_FLAG_ANYSRC
: 0;
206 void ip_send_unicast_reply(struct sock
*sk
, struct sk_buff
*skb
,
207 const struct ip_options
*sopt
,
208 __be32 daddr
, __be32 saddr
,
209 const struct ip_reply_arg
*arg
,
212 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
213 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
214 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
215 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
216 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
217 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
218 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
219 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
220 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
221 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
223 u64
snmp_get_cpu_field(void __percpu
*mib
, int cpu
, int offct
);
224 unsigned long snmp_fold_field(void __percpu
*mib
, int offt
);
225 #if BITS_PER_LONG==32
226 u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
227 size_t syncp_offset
);
228 u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t sync_off
);
230 static inline u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
233 return snmp_get_cpu_field(mib
, cpu
, offct
);
237 static inline u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t syncp_off
)
239 return snmp_fold_field(mib
, offt
);
243 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
246 for_each_possible_cpu(c) { \
247 for (i = 0; stats_list[i].name; i++) \
248 buff64[i] += snmp_get_cpu_field64( \
250 c, stats_list[i].entry, \
255 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
258 for_each_possible_cpu(c) { \
259 for (i = 0; stats_list[i].name; i++) \
260 buff[i] += snmp_get_cpu_field( \
262 c, stats_list[i].entry); \
266 void inet_get_local_port_range(struct net
*net
, int *low
, int *high
);
269 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
271 if (!net
->ipv4
.sysctl_local_reserved_ports
)
273 return test_bit(port
, net
->ipv4
.sysctl_local_reserved_ports
);
276 static inline bool sysctl_dev_name_is_allowed(const char *name
)
278 return strcmp(name
, "default") != 0 && strcmp(name
, "all") != 0;
281 static inline int inet_prot_sock(struct net
*net
)
283 return net
->ipv4
.sysctl_ip_prot_sock
;
287 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
292 static inline int inet_prot_sock(struct net
*net
)
298 __be32
inet_current_timestamp(void);
300 /* From inetpeer.c */
301 extern int inet_peer_threshold
;
302 extern int inet_peer_minttl
;
303 extern int inet_peer_maxttl
;
305 void ipfrag_init(void);
307 void ip_static_sysctl_init(void);
309 #define IP4_REPLY_MARK(net, mark) \
310 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
312 static inline bool ip_is_fragment(const struct iphdr
*iph
)
314 return (iph
->frag_off
& htons(IP_MF
| IP_OFFSET
)) != 0;
320 /* The function in 2.2 was invalid, producing wrong result for
321 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
323 int ip_decrease_ttl(struct iphdr
*iph
)
325 u32 check
= (__force u32
)iph
->check
;
326 check
+= (__force u32
)htons(0x0100);
327 iph
->check
= (__force __sum16
)(check
+ (check
>=0xFFFF));
331 static inline int ip_mtu_locked(const struct dst_entry
*dst
)
333 const struct rtable
*rt
= (const struct rtable
*)dst
;
335 return rt
->rt_mtu_locked
|| dst_metric_locked(dst
, RTAX_MTU
);
339 int ip_dont_fragment(const struct sock
*sk
, const struct dst_entry
*dst
)
341 u8 pmtudisc
= READ_ONCE(inet_sk(sk
)->pmtudisc
);
343 return pmtudisc
== IP_PMTUDISC_DO
||
344 (pmtudisc
== IP_PMTUDISC_WANT
&&
345 !ip_mtu_locked(dst
));
348 static inline bool ip_sk_accept_pmtu(const struct sock
*sk
)
350 return inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_INTERFACE
&&
351 inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_OMIT
;
354 static inline bool ip_sk_use_pmtu(const struct sock
*sk
)
356 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_PROBE
;
359 static inline bool ip_sk_ignore_df(const struct sock
*sk
)
361 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_DO
||
362 inet_sk(sk
)->pmtudisc
== IP_PMTUDISC_OMIT
;
365 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry
*dst
,
368 struct net
*net
= dev_net(dst
->dev
);
370 if (net
->ipv4
.sysctl_ip_fwd_use_pmtu
||
371 ip_mtu_locked(dst
) ||
375 return min(READ_ONCE(dst
->dev
->mtu
), IP_MAX_MTU
);
378 static inline unsigned int ip_skb_dst_mtu(struct sock
*sk
,
379 const struct sk_buff
*skb
)
381 if (!sk
|| !sk_fullsock(sk
) || ip_sk_use_pmtu(sk
)) {
382 bool forwarding
= IPCB(skb
)->flags
& IPSKB_FORWARDED
;
384 return ip_dst_mtu_maybe_forward(skb_dst(skb
), forwarding
);
387 return min(READ_ONCE(skb_dst(skb
)->dev
->mtu
), IP_MAX_MTU
);
390 u32
ip_idents_reserve(u32 hash
, int segs
);
391 void __ip_select_ident(struct net
*net
, struct iphdr
*iph
, int segs
);
393 static inline void ip_select_ident_segs(struct net
*net
, struct sk_buff
*skb
,
394 struct sock
*sk
, int segs
)
396 struct iphdr
*iph
= ip_hdr(skb
);
398 if ((iph
->frag_off
& htons(IP_DF
)) && !skb
->ignore_df
) {
399 /* This is only to work around buggy Windows95/2000
400 * VJ compression implementations. If the ID field
401 * does not change, they drop every other packet in
402 * a TCP stream using header compression.
404 if (sk
&& inet_sk(sk
)->inet_daddr
) {
405 iph
->id
= htons(inet_sk(sk
)->inet_id
);
406 inet_sk(sk
)->inet_id
+= segs
;
411 __ip_select_ident(net
, iph
, segs
);
415 static inline void ip_select_ident(struct net
*net
, struct sk_buff
*skb
,
418 ip_select_ident_segs(net
, skb
, sk
, 1);
421 static inline __wsum
inet_compute_pseudo(struct sk_buff
*skb
, int proto
)
423 return csum_tcpudp_nofold(ip_hdr(skb
)->saddr
, ip_hdr(skb
)->daddr
,
427 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
428 * Equivalent to : flow->v4addrs.src = iph->saddr;
429 * flow->v4addrs.dst = iph->daddr;
431 static inline void iph_to_flow_copy_v4addrs(struct flow_keys
*flow
,
432 const struct iphdr
*iph
)
434 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v4addrs
.dst
) !=
435 offsetof(typeof(flow
->addrs
), v4addrs
.src
) +
436 sizeof(flow
->addrs
.v4addrs
.src
));
437 memcpy(&flow
->addrs
.v4addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v4addrs
));
438 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV4_ADDRS
;
441 static inline __wsum
inet_gro_compute_pseudo(struct sk_buff
*skb
, int proto
)
443 const struct iphdr
*iph
= skb_gro_network_header(skb
);
445 return csum_tcpudp_nofold(iph
->saddr
, iph
->daddr
,
446 skb_gro_len(skb
), proto
, 0);
450 * Map a multicast IP onto multicast MAC for type ethernet.
453 static inline void ip_eth_mc_map(__be32 naddr
, char *buf
)
455 __u32 addr
=ntohl(naddr
);
467 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
468 * Leave P_Key as 0 to be filled in by driver.
471 static inline void ip_ib_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
474 unsigned char scope
= broadcast
[5] & 0xF;
476 buf
[0] = 0; /* Reserved */
477 buf
[1] = 0xff; /* Multicast QPN */
482 buf
[5] = 0x10 | scope
; /* scope from broadcast address */
483 buf
[6] = 0x40; /* IPv4 signature */
485 buf
[8] = broadcast
[8]; /* P_Key */
486 buf
[9] = broadcast
[9];
493 buf
[19] = addr
& 0xff;
495 buf
[18] = addr
& 0xff;
497 buf
[17] = addr
& 0xff;
499 buf
[16] = addr
& 0x0f;
502 static inline void ip_ipgre_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
504 if ((broadcast
[0] | broadcast
[1] | broadcast
[2] | broadcast
[3]) != 0)
505 memcpy(buf
, broadcast
, 4);
507 memcpy(buf
, &naddr
, sizeof(naddr
));
510 #if IS_ENABLED(CONFIG_IPV6)
511 #include <linux/ipv6.h>
514 static __inline__
void inet_reset_saddr(struct sock
*sk
)
516 inet_sk(sk
)->inet_rcv_saddr
= inet_sk(sk
)->inet_saddr
= 0;
517 #if IS_ENABLED(CONFIG_IPV6)
518 if (sk
->sk_family
== PF_INET6
) {
519 struct ipv6_pinfo
*np
= inet6_sk(sk
);
521 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
522 memset(&sk
->sk_v6_rcv_saddr
, 0, sizeof(sk
->sk_v6_rcv_saddr
));
529 static inline unsigned int ipv4_addr_hash(__be32 ip
)
531 return (__force
unsigned int) ip
;
534 static inline u32
ipv4_portaddr_hash(const struct net
*net
,
538 return jhash_1word((__force u32
)saddr
, net_hash_mix(net
)) ^ port
;
541 bool ip_call_ra_chain(struct sk_buff
*skb
);
544 * Functions provided by ip_fragment.c
547 enum ip_defrag_users
{
548 IP_DEFRAG_LOCAL_DELIVER
,
549 IP_DEFRAG_CALL_RA_CHAIN
,
550 IP_DEFRAG_CONNTRACK_IN
,
551 __IP_DEFRAG_CONNTRACK_IN_END
= IP_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
552 IP_DEFRAG_CONNTRACK_OUT
,
553 __IP_DEFRAG_CONNTRACK_OUT_END
= IP_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
554 IP_DEFRAG_CONNTRACK_BRIDGE_IN
,
555 __IP_DEFRAG_CONNTRACK_BRIDGE_IN
= IP_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
563 /* Return true if the value of 'user' is between 'lower_bond'
564 * and 'upper_bond' inclusively.
566 static inline bool ip_defrag_user_in_between(u32 user
,
567 enum ip_defrag_users lower_bond
,
568 enum ip_defrag_users upper_bond
)
570 return user
>= lower_bond
&& user
<= upper_bond
;
573 int ip_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
575 struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
577 static inline struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
)
582 int ip_frag_mem(struct net
*net
);
585 * Functions provided by ip_forward.c
588 int ip_forward(struct sk_buff
*skb
);
591 * Functions provided by ip_options.c
594 void ip_options_build(struct sk_buff
*skb
, struct ip_options
*opt
,
595 __be32 daddr
, struct rtable
*rt
, int is_frag
);
597 int __ip_options_echo(struct net
*net
, struct ip_options
*dopt
,
598 struct sk_buff
*skb
, const struct ip_options
*sopt
);
599 static inline int ip_options_echo(struct net
*net
, struct ip_options
*dopt
,
602 return __ip_options_echo(net
, dopt
, skb
, &IPCB(skb
)->opt
);
605 void ip_options_fragment(struct sk_buff
*skb
);
606 int ip_options_compile(struct net
*net
, struct ip_options
*opt
,
607 struct sk_buff
*skb
);
608 int ip_options_get(struct net
*net
, struct ip_options_rcu
**optp
,
609 unsigned char *data
, int optlen
);
610 int ip_options_get_from_user(struct net
*net
, struct ip_options_rcu
**optp
,
611 unsigned char __user
*data
, int optlen
);
612 void ip_options_undo(struct ip_options
*opt
);
613 void ip_forward_options(struct sk_buff
*skb
);
614 int ip_options_rcv_srr(struct sk_buff
*skb
);
617 * Functions provided by ip_sockglue.c
620 void ipv4_pktinfo_prepare(const struct sock
*sk
, struct sk_buff
*skb
);
621 void ip_cmsg_recv_offset(struct msghdr
*msg
, struct sock
*sk
,
622 struct sk_buff
*skb
, int tlen
, int offset
);
623 int ip_cmsg_send(struct sock
*sk
, struct msghdr
*msg
,
624 struct ipcm_cookie
*ipc
, bool allow_ipv6
);
625 int ip_setsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
626 unsigned int optlen
);
627 int ip_getsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
629 int compat_ip_setsockopt(struct sock
*sk
, int level
, int optname
,
630 char __user
*optval
, unsigned int optlen
);
631 int compat_ip_getsockopt(struct sock
*sk
, int level
, int optname
,
632 char __user
*optval
, int __user
*optlen
);
633 int ip_ra_control(struct sock
*sk
, unsigned char on
,
634 void (*destructor
)(struct sock
*));
636 int ip_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
, int *addr_len
);
637 void ip_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
638 u32 info
, u8
*payload
);
639 void ip_local_error(struct sock
*sk
, int err
, __be32 daddr
, __be16 dport
,
642 static inline void ip_cmsg_recv(struct msghdr
*msg
, struct sk_buff
*skb
)
644 ip_cmsg_recv_offset(msg
, skb
->sk
, skb
, 0, 0);
647 bool icmp_global_allow(void);
648 extern int sysctl_icmp_msgs_per_sec
;
649 extern int sysctl_icmp_msgs_burst
;
651 #ifdef CONFIG_PROC_FS
652 int ip_misc_proc_init(void);