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_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */
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))
81 /* return enslaved device index if relevant */
82 static inline int inet_sdif(struct sk_buff
*skb
)
84 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
85 if (skb
&& ipv4_l3mdev_skb(IPCB(skb
)->flags
))
86 return IPCB(skb
)->iif
;
92 struct ip_ra_chain __rcu
*next
;
95 void (*destructor
)(struct sock
*);
96 struct sock
*saved_sk
;
101 extern struct ip_ra_chain __rcu
*ip_ra_chain
;
104 #define IP_CE 0x8000 /* Flag: "Congestion" */
105 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
106 #define IP_MF 0x2000 /* Flag: "More Fragments" */
107 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
109 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
117 int igmp_mc_init(void);
120 * Functions provided by ip.c
123 int ip_build_and_send_pkt(struct sk_buff
*skb
, const struct sock
*sk
,
124 __be32 saddr
, __be32 daddr
,
125 struct ip_options_rcu
*opt
);
126 int ip_rcv(struct sk_buff
*skb
, struct net_device
*dev
, struct packet_type
*pt
,
127 struct net_device
*orig_dev
);
128 int ip_local_deliver(struct sk_buff
*skb
);
129 int ip_mr_input(struct sk_buff
*skb
);
130 int ip_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
131 int ip_mc_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
132 int ip_do_fragment(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
,
133 int (*output
)(struct net
*, struct sock
*, struct sk_buff
*));
134 void ip_send_check(struct iphdr
*ip
);
135 int __ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
136 int ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
138 int ip_queue_xmit(struct sock
*sk
, struct sk_buff
*skb
, struct flowi
*fl
);
140 int ip_append_data(struct sock
*sk
, struct flowi4
*fl4
,
141 int getfrag(void *from
, char *to
, int offset
, int len
,
142 int odd
, struct sk_buff
*skb
),
143 void *from
, int len
, int protolen
,
144 struct ipcm_cookie
*ipc
,
147 int ip_generic_getfrag(void *from
, char *to
, int offset
, int len
, int odd
,
148 struct sk_buff
*skb
);
149 ssize_t
ip_append_page(struct sock
*sk
, struct flowi4
*fl4
, struct page
*page
,
150 int offset
, size_t size
, int flags
);
151 struct sk_buff
*__ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
152 struct sk_buff_head
*queue
,
153 struct inet_cork
*cork
);
154 int ip_send_skb(struct net
*net
, struct sk_buff
*skb
);
155 int ip_push_pending_frames(struct sock
*sk
, struct flowi4
*fl4
);
156 void ip_flush_pending_frames(struct sock
*sk
);
157 struct sk_buff
*ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
158 int getfrag(void *from
, char *to
, int offset
,
159 int len
, int odd
, struct sk_buff
*skb
),
160 void *from
, int length
, int transhdrlen
,
161 struct ipcm_cookie
*ipc
, struct rtable
**rtp
,
164 static inline struct sk_buff
*ip_finish_skb(struct sock
*sk
, struct flowi4
*fl4
)
166 return __ip_make_skb(sk
, fl4
, &sk
->sk_write_queue
, &inet_sk(sk
)->cork
.base
);
169 static inline __u8
get_rttos(struct ipcm_cookie
* ipc
, struct inet_sock
*inet
)
171 return (ipc
->tos
!= -1) ? RT_TOS(ipc
->tos
) : RT_TOS(inet
->tos
);
174 static inline __u8
get_rtconn_flags(struct ipcm_cookie
* ipc
, struct sock
* sk
)
176 return (ipc
->tos
!= -1) ? RT_CONN_FLAGS_TOS(sk
, ipc
->tos
) : RT_CONN_FLAGS(sk
);
180 int __ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
181 int ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
183 void ip4_datagram_release_cb(struct sock
*sk
);
185 struct ip_reply_arg
{
189 int csumoffset
; /* u16 offset of csum in iov[0].iov_base */
190 /* -1 if not needed */
196 #define IP_REPLY_ARG_NOSRCCHECK 1
198 static inline __u8
ip_reply_arg_flowi_flags(const struct ip_reply_arg
*arg
)
200 return (arg
->flags
& IP_REPLY_ARG_NOSRCCHECK
) ? FLOWI_FLAG_ANYSRC
: 0;
203 void ip_send_unicast_reply(struct sock
*sk
, struct sk_buff
*skb
,
204 const struct ip_options
*sopt
,
205 __be32 daddr
, __be32 saddr
,
206 const struct ip_reply_arg
*arg
,
209 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
210 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
211 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
212 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
213 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
214 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
215 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
216 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
217 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
218 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
220 u64
snmp_get_cpu_field(void __percpu
*mib
, int cpu
, int offct
);
221 unsigned long snmp_fold_field(void __percpu
*mib
, int offt
);
222 #if BITS_PER_LONG==32
223 u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
224 size_t syncp_offset
);
225 u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t sync_off
);
227 static inline u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
230 return snmp_get_cpu_field(mib
, cpu
, offct
);
234 static inline u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t syncp_off
)
236 return snmp_fold_field(mib
, offt
);
240 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
243 for_each_possible_cpu(c) { \
244 for (i = 0; stats_list[i].name; i++) \
245 buff64[i] += snmp_get_cpu_field64( \
247 c, stats_list[i].entry, \
252 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
255 for_each_possible_cpu(c) { \
256 for (i = 0; stats_list[i].name; i++) \
257 buff[i] += snmp_get_cpu_field( \
259 c, stats_list[i].entry); \
263 void inet_get_local_port_range(struct net
*net
, int *low
, int *high
);
266 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
268 if (!net
->ipv4
.sysctl_local_reserved_ports
)
270 return test_bit(port
, net
->ipv4
.sysctl_local_reserved_ports
);
273 static inline bool sysctl_dev_name_is_allowed(const char *name
)
275 return strcmp(name
, "default") != 0 && strcmp(name
, "all") != 0;
278 static inline int inet_prot_sock(struct net
*net
)
280 return net
->ipv4
.sysctl_ip_prot_sock
;
284 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
289 static inline int inet_prot_sock(struct net
*net
)
295 __be32
inet_current_timestamp(void);
297 /* From inetpeer.c */
298 extern int inet_peer_threshold
;
299 extern int inet_peer_minttl
;
300 extern int inet_peer_maxttl
;
302 void ipfrag_init(void);
304 void ip_static_sysctl_init(void);
306 #define IP4_REPLY_MARK(net, mark) \
307 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
309 static inline bool ip_is_fragment(const struct iphdr
*iph
)
311 return (iph
->frag_off
& htons(IP_MF
| IP_OFFSET
)) != 0;
317 /* The function in 2.2 was invalid, producing wrong result for
318 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
320 int ip_decrease_ttl(struct iphdr
*iph
)
322 u32 check
= (__force u32
)iph
->check
;
323 check
+= (__force u32
)htons(0x0100);
324 iph
->check
= (__force __sum16
)(check
+ (check
>=0xFFFF));
329 int ip_dont_fragment(const struct sock
*sk
, const struct dst_entry
*dst
)
331 u8 pmtudisc
= READ_ONCE(inet_sk(sk
)->pmtudisc
);
333 return pmtudisc
== IP_PMTUDISC_DO
||
334 (pmtudisc
== IP_PMTUDISC_WANT
&&
335 !(dst_metric_locked(dst
, RTAX_MTU
)));
338 static inline bool ip_sk_accept_pmtu(const struct sock
*sk
)
340 return inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_INTERFACE
&&
341 inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_OMIT
;
344 static inline bool ip_sk_use_pmtu(const struct sock
*sk
)
346 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_PROBE
;
349 static inline bool ip_sk_ignore_df(const struct sock
*sk
)
351 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_DO
||
352 inet_sk(sk
)->pmtudisc
== IP_PMTUDISC_OMIT
;
355 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry
*dst
,
358 struct net
*net
= dev_net(dst
->dev
);
360 if (net
->ipv4
.sysctl_ip_fwd_use_pmtu
||
361 dst_metric_locked(dst
, RTAX_MTU
) ||
365 return min(READ_ONCE(dst
->dev
->mtu
), IP_MAX_MTU
);
368 static inline unsigned int ip_skb_dst_mtu(struct sock
*sk
,
369 const struct sk_buff
*skb
)
371 if (!sk
|| !sk_fullsock(sk
) || ip_sk_use_pmtu(sk
)) {
372 bool forwarding
= IPCB(skb
)->flags
& IPSKB_FORWARDED
;
374 return ip_dst_mtu_maybe_forward(skb_dst(skb
), forwarding
);
377 return min(READ_ONCE(skb_dst(skb
)->dev
->mtu
), IP_MAX_MTU
);
380 u32
ip_idents_reserve(u32 hash
, int segs
);
381 void __ip_select_ident(struct net
*net
, struct iphdr
*iph
, int segs
);
383 static inline void ip_select_ident_segs(struct net
*net
, struct sk_buff
*skb
,
384 struct sock
*sk
, int segs
)
386 struct iphdr
*iph
= ip_hdr(skb
);
388 if ((iph
->frag_off
& htons(IP_DF
)) && !skb
->ignore_df
) {
389 /* This is only to work around buggy Windows95/2000
390 * VJ compression implementations. If the ID field
391 * does not change, they drop every other packet in
392 * a TCP stream using header compression.
394 if (sk
&& inet_sk(sk
)->inet_daddr
) {
395 iph
->id
= htons(inet_sk(sk
)->inet_id
);
396 inet_sk(sk
)->inet_id
+= segs
;
401 __ip_select_ident(net
, iph
, segs
);
405 static inline void ip_select_ident(struct net
*net
, struct sk_buff
*skb
,
408 ip_select_ident_segs(net
, skb
, sk
, 1);
411 static inline __wsum
inet_compute_pseudo(struct sk_buff
*skb
, int proto
)
413 return csum_tcpudp_nofold(ip_hdr(skb
)->saddr
, ip_hdr(skb
)->daddr
,
417 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
418 * Equivalent to : flow->v4addrs.src = iph->saddr;
419 * flow->v4addrs.dst = iph->daddr;
421 static inline void iph_to_flow_copy_v4addrs(struct flow_keys
*flow
,
422 const struct iphdr
*iph
)
424 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v4addrs
.dst
) !=
425 offsetof(typeof(flow
->addrs
), v4addrs
.src
) +
426 sizeof(flow
->addrs
.v4addrs
.src
));
427 memcpy(&flow
->addrs
.v4addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v4addrs
));
428 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV4_ADDRS
;
431 static inline __wsum
inet_gro_compute_pseudo(struct sk_buff
*skb
, int proto
)
433 const struct iphdr
*iph
= skb_gro_network_header(skb
);
435 return csum_tcpudp_nofold(iph
->saddr
, iph
->daddr
,
436 skb_gro_len(skb
), proto
, 0);
440 * Map a multicast IP onto multicast MAC for type ethernet.
443 static inline void ip_eth_mc_map(__be32 naddr
, char *buf
)
445 __u32 addr
=ntohl(naddr
);
457 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
458 * Leave P_Key as 0 to be filled in by driver.
461 static inline void ip_ib_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
464 unsigned char scope
= broadcast
[5] & 0xF;
466 buf
[0] = 0; /* Reserved */
467 buf
[1] = 0xff; /* Multicast QPN */
472 buf
[5] = 0x10 | scope
; /* scope from broadcast address */
473 buf
[6] = 0x40; /* IPv4 signature */
475 buf
[8] = broadcast
[8]; /* P_Key */
476 buf
[9] = broadcast
[9];
483 buf
[19] = addr
& 0xff;
485 buf
[18] = addr
& 0xff;
487 buf
[17] = addr
& 0xff;
489 buf
[16] = addr
& 0x0f;
492 static inline void ip_ipgre_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
494 if ((broadcast
[0] | broadcast
[1] | broadcast
[2] | broadcast
[3]) != 0)
495 memcpy(buf
, broadcast
, 4);
497 memcpy(buf
, &naddr
, sizeof(naddr
));
500 #if IS_ENABLED(CONFIG_IPV6)
501 #include <linux/ipv6.h>
504 static __inline__
void inet_reset_saddr(struct sock
*sk
)
506 inet_sk(sk
)->inet_rcv_saddr
= inet_sk(sk
)->inet_saddr
= 0;
507 #if IS_ENABLED(CONFIG_IPV6)
508 if (sk
->sk_family
== PF_INET6
) {
509 struct ipv6_pinfo
*np
= inet6_sk(sk
);
511 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
512 memset(&sk
->sk_v6_rcv_saddr
, 0, sizeof(sk
->sk_v6_rcv_saddr
));
519 static inline unsigned int ipv4_addr_hash(__be32 ip
)
521 return (__force
unsigned int) ip
;
524 bool ip_call_ra_chain(struct sk_buff
*skb
);
527 * Functions provided by ip_fragment.c
530 enum ip_defrag_users
{
531 IP_DEFRAG_LOCAL_DELIVER
,
532 IP_DEFRAG_CALL_RA_CHAIN
,
533 IP_DEFRAG_CONNTRACK_IN
,
534 __IP_DEFRAG_CONNTRACK_IN_END
= IP_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
535 IP_DEFRAG_CONNTRACK_OUT
,
536 __IP_DEFRAG_CONNTRACK_OUT_END
= IP_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
537 IP_DEFRAG_CONNTRACK_BRIDGE_IN
,
538 __IP_DEFRAG_CONNTRACK_BRIDGE_IN
= IP_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
546 /* Return true if the value of 'user' is between 'lower_bond'
547 * and 'upper_bond' inclusively.
549 static inline bool ip_defrag_user_in_between(u32 user
,
550 enum ip_defrag_users lower_bond
,
551 enum ip_defrag_users upper_bond
)
553 return user
>= lower_bond
&& user
<= upper_bond
;
556 int ip_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
558 struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
560 static inline struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
)
565 int ip_frag_mem(struct net
*net
);
568 * Functions provided by ip_forward.c
571 int ip_forward(struct sk_buff
*skb
);
574 * Functions provided by ip_options.c
577 void ip_options_build(struct sk_buff
*skb
, struct ip_options
*opt
,
578 __be32 daddr
, struct rtable
*rt
, int is_frag
);
580 int __ip_options_echo(struct net
*net
, struct ip_options
*dopt
,
581 struct sk_buff
*skb
, const struct ip_options
*sopt
);
582 static inline int ip_options_echo(struct net
*net
, struct ip_options
*dopt
,
585 return __ip_options_echo(net
, dopt
, skb
, &IPCB(skb
)->opt
);
588 void ip_options_fragment(struct sk_buff
*skb
);
589 int ip_options_compile(struct net
*net
, struct ip_options
*opt
,
590 struct sk_buff
*skb
);
591 int ip_options_get(struct net
*net
, struct ip_options_rcu
**optp
,
592 unsigned char *data
, int optlen
);
593 int ip_options_get_from_user(struct net
*net
, struct ip_options_rcu
**optp
,
594 unsigned char __user
*data
, int optlen
);
595 void ip_options_undo(struct ip_options
*opt
);
596 void ip_forward_options(struct sk_buff
*skb
);
597 int ip_options_rcv_srr(struct sk_buff
*skb
);
600 * Functions provided by ip_sockglue.c
603 void ipv4_pktinfo_prepare(const struct sock
*sk
, struct sk_buff
*skb
);
604 void ip_cmsg_recv_offset(struct msghdr
*msg
, struct sock
*sk
,
605 struct sk_buff
*skb
, int tlen
, int offset
);
606 int ip_cmsg_send(struct sock
*sk
, struct msghdr
*msg
,
607 struct ipcm_cookie
*ipc
, bool allow_ipv6
);
608 int ip_setsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
609 unsigned int optlen
);
610 int ip_getsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
612 int compat_ip_setsockopt(struct sock
*sk
, int level
, int optname
,
613 char __user
*optval
, unsigned int optlen
);
614 int compat_ip_getsockopt(struct sock
*sk
, int level
, int optname
,
615 char __user
*optval
, int __user
*optlen
);
616 int ip_ra_control(struct sock
*sk
, unsigned char on
,
617 void (*destructor
)(struct sock
*));
619 int ip_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
, int *addr_len
);
620 void ip_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
621 u32 info
, u8
*payload
);
622 void ip_local_error(struct sock
*sk
, int err
, __be32 daddr
, __be16 dport
,
625 static inline void ip_cmsg_recv(struct msghdr
*msg
, struct sk_buff
*skb
)
627 ip_cmsg_recv_offset(msg
, skb
->sk
, skb
, 0, 0);
630 bool icmp_global_allow(void);
631 extern int sysctl_icmp_msgs_per_sec
;
632 extern int sysctl_icmp_msgs_burst
;
634 #ifdef CONFIG_PROC_FS
635 int ip_misc_proc_init(void);