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>
38 struct inet_skb_parm
{
39 struct ip_options opt
; /* Compiled IP options */
42 #define IPSKB_FORWARDED BIT(0)
43 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
44 #define IPSKB_XFRM_TRANSFORMED BIT(2)
45 #define IPSKB_FRAG_COMPLETE BIT(3)
46 #define IPSKB_REROUTED BIT(4)
47 #define IPSKB_DOREDIRECT BIT(5)
48 #define IPSKB_FRAG_PMTU BIT(6)
53 static inline unsigned int ip_hdrlen(const struct sk_buff
*skb
)
55 return ip_hdr(skb
)->ihl
* 4;
59 struct sockcm_cookie sockc
;
62 struct ip_options_rcu
*opt
;
69 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
70 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
73 struct ip_ra_chain __rcu
*next
;
76 void (*destructor
)(struct sock
*);
77 struct sock
*saved_sk
;
82 extern struct ip_ra_chain __rcu
*ip_ra_chain
;
85 #define IP_CE 0x8000 /* Flag: "Congestion" */
86 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
87 #define IP_MF 0x2000 /* Flag: "More Fragments" */
88 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
90 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
98 int igmp_mc_init(void);
101 * Functions provided by ip.c
104 int ip_build_and_send_pkt(struct sk_buff
*skb
, const struct sock
*sk
,
105 __be32 saddr
, __be32 daddr
,
106 struct ip_options_rcu
*opt
);
107 int ip_rcv(struct sk_buff
*skb
, struct net_device
*dev
, struct packet_type
*pt
,
108 struct net_device
*orig_dev
);
109 int ip_local_deliver(struct sk_buff
*skb
);
110 int ip_mr_input(struct sk_buff
*skb
);
111 int ip_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
112 int ip_mc_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
113 int ip_do_fragment(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
,
114 int (*output
)(struct net
*, struct sock
*, struct sk_buff
*));
115 void ip_send_check(struct iphdr
*ip
);
116 int __ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
117 int ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
119 int ip_queue_xmit(struct sock
*sk
, struct sk_buff
*skb
, struct flowi
*fl
);
121 int ip_append_data(struct sock
*sk
, struct flowi4
*fl4
,
122 int getfrag(void *from
, char *to
, int offset
, int len
,
123 int odd
, struct sk_buff
*skb
),
124 void *from
, int len
, int protolen
,
125 struct ipcm_cookie
*ipc
,
128 int ip_generic_getfrag(void *from
, char *to
, int offset
, int len
, int odd
,
129 struct sk_buff
*skb
);
130 ssize_t
ip_append_page(struct sock
*sk
, struct flowi4
*fl4
, struct page
*page
,
131 int offset
, size_t size
, int flags
);
132 struct sk_buff
*__ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
133 struct sk_buff_head
*queue
,
134 struct inet_cork
*cork
);
135 int ip_send_skb(struct net
*net
, struct sk_buff
*skb
);
136 int ip_push_pending_frames(struct sock
*sk
, struct flowi4
*fl4
);
137 void ip_flush_pending_frames(struct sock
*sk
);
138 struct sk_buff
*ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
139 int getfrag(void *from
, char *to
, int offset
,
140 int len
, int odd
, struct sk_buff
*skb
),
141 void *from
, int length
, int transhdrlen
,
142 struct ipcm_cookie
*ipc
, struct rtable
**rtp
,
145 static inline struct sk_buff
*ip_finish_skb(struct sock
*sk
, struct flowi4
*fl4
)
147 return __ip_make_skb(sk
, fl4
, &sk
->sk_write_queue
, &inet_sk(sk
)->cork
.base
);
150 static inline __u8
get_rttos(struct ipcm_cookie
* ipc
, struct inet_sock
*inet
)
152 return (ipc
->tos
!= -1) ? RT_TOS(ipc
->tos
) : RT_TOS(inet
->tos
);
155 static inline __u8
get_rtconn_flags(struct ipcm_cookie
* ipc
, struct sock
* sk
)
157 return (ipc
->tos
!= -1) ? RT_CONN_FLAGS_TOS(sk
, ipc
->tos
) : RT_CONN_FLAGS(sk
);
161 int __ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
162 int ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
164 void ip4_datagram_release_cb(struct sock
*sk
);
166 struct ip_reply_arg
{
170 int csumoffset
; /* u16 offset of csum in iov[0].iov_base */
171 /* -1 if not needed */
176 #define IP_REPLY_ARG_NOSRCCHECK 1
178 static inline __u8
ip_reply_arg_flowi_flags(const struct ip_reply_arg
*arg
)
180 return (arg
->flags
& IP_REPLY_ARG_NOSRCCHECK
) ? FLOWI_FLAG_ANYSRC
: 0;
183 void ip_send_unicast_reply(struct sock
*sk
, struct sk_buff
*skb
,
184 const struct ip_options
*sopt
,
185 __be32 daddr
, __be32 saddr
,
186 const struct ip_reply_arg
*arg
,
189 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
190 #define IP_INC_STATS_BH(net, field) SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
191 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
192 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
193 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
194 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
195 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
196 #define NET_INC_STATS_BH(net, field) SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
197 #define NET_INC_STATS_USER(net, field) SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
198 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
199 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
200 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
202 u64
snmp_get_cpu_field(void __percpu
*mib
, int cpu
, int offct
);
203 unsigned long snmp_fold_field(void __percpu
*mib
, int offt
);
204 #if BITS_PER_LONG==32
205 u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
206 size_t syncp_offset
);
207 u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t sync_off
);
209 static inline u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
212 return snmp_get_cpu_field(mib
, cpu
, offct
);
216 static inline u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t syncp_off
)
218 return snmp_fold_field(mib
, offt
);
222 void inet_get_local_port_range(struct net
*net
, int *low
, int *high
);
225 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
227 if (!net
->ipv4
.sysctl_local_reserved_ports
)
229 return test_bit(port
, net
->ipv4
.sysctl_local_reserved_ports
);
232 static inline bool sysctl_dev_name_is_allowed(const char *name
)
234 return strcmp(name
, "default") != 0 && strcmp(name
, "all") != 0;
238 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
244 __be32
inet_current_timestamp(void);
246 /* From inetpeer.c */
247 extern int inet_peer_threshold
;
248 extern int inet_peer_minttl
;
249 extern int inet_peer_maxttl
;
251 void ipfrag_init(void);
253 void ip_static_sysctl_init(void);
255 #define IP4_REPLY_MARK(net, mark) \
256 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
258 static inline bool ip_is_fragment(const struct iphdr
*iph
)
260 return (iph
->frag_off
& htons(IP_MF
| IP_OFFSET
)) != 0;
266 /* The function in 2.2 was invalid, producing wrong result for
267 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
269 int ip_decrease_ttl(struct iphdr
*iph
)
271 u32 check
= (__force u32
)iph
->check
;
272 check
+= (__force u32
)htons(0x0100);
273 iph
->check
= (__force __sum16
)(check
+ (check
>=0xFFFF));
278 int ip_dont_fragment(const struct sock
*sk
, const struct dst_entry
*dst
)
280 u8 pmtudisc
= READ_ONCE(inet_sk(sk
)->pmtudisc
);
282 return pmtudisc
== IP_PMTUDISC_DO
||
283 (pmtudisc
== IP_PMTUDISC_WANT
&&
284 !(dst_metric_locked(dst
, RTAX_MTU
)));
287 static inline bool ip_sk_accept_pmtu(const struct sock
*sk
)
289 return inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_INTERFACE
&&
290 inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_OMIT
;
293 static inline bool ip_sk_use_pmtu(const struct sock
*sk
)
295 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_PROBE
;
298 static inline bool ip_sk_ignore_df(const struct sock
*sk
)
300 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_DO
||
301 inet_sk(sk
)->pmtudisc
== IP_PMTUDISC_OMIT
;
304 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry
*dst
,
307 struct net
*net
= dev_net(dst
->dev
);
309 if (net
->ipv4
.sysctl_ip_fwd_use_pmtu
||
310 dst_metric_locked(dst
, RTAX_MTU
) ||
314 return min(dst
->dev
->mtu
, IP_MAX_MTU
);
317 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff
*skb
)
319 struct sock
*sk
= skb
->sk
;
321 if (!sk
|| !sk_fullsock(sk
) || ip_sk_use_pmtu(sk
)) {
322 bool forwarding
= IPCB(skb
)->flags
& IPSKB_FORWARDED
;
324 return ip_dst_mtu_maybe_forward(skb_dst(skb
), forwarding
);
327 return min(skb_dst(skb
)->dev
->mtu
, IP_MAX_MTU
);
330 u32
ip_idents_reserve(u32 hash
, int segs
);
331 void __ip_select_ident(struct net
*net
, struct iphdr
*iph
, int segs
);
333 static inline void ip_select_ident_segs(struct net
*net
, struct sk_buff
*skb
,
334 struct sock
*sk
, int segs
)
336 struct iphdr
*iph
= ip_hdr(skb
);
338 if ((iph
->frag_off
& htons(IP_DF
)) && !skb
->ignore_df
) {
339 /* This is only to work around buggy Windows95/2000
340 * VJ compression implementations. If the ID field
341 * does not change, they drop every other packet in
342 * a TCP stream using header compression.
344 if (sk
&& inet_sk(sk
)->inet_daddr
) {
345 iph
->id
= htons(inet_sk(sk
)->inet_id
);
346 inet_sk(sk
)->inet_id
+= segs
;
351 __ip_select_ident(net
, iph
, segs
);
355 static inline void ip_select_ident(struct net
*net
, struct sk_buff
*skb
,
358 ip_select_ident_segs(net
, skb
, sk
, 1);
361 static inline __wsum
inet_compute_pseudo(struct sk_buff
*skb
, int proto
)
363 return csum_tcpudp_nofold(ip_hdr(skb
)->saddr
, ip_hdr(skb
)->daddr
,
367 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
368 * Equivalent to : flow->v4addrs.src = iph->saddr;
369 * flow->v4addrs.dst = iph->daddr;
371 static inline void iph_to_flow_copy_v4addrs(struct flow_keys
*flow
,
372 const struct iphdr
*iph
)
374 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v4addrs
.dst
) !=
375 offsetof(typeof(flow
->addrs
), v4addrs
.src
) +
376 sizeof(flow
->addrs
.v4addrs
.src
));
377 memcpy(&flow
->addrs
.v4addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v4addrs
));
378 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV4_ADDRS
;
381 static inline __wsum
inet_gro_compute_pseudo(struct sk_buff
*skb
, int proto
)
383 const struct iphdr
*iph
= skb_gro_network_header(skb
);
385 return csum_tcpudp_nofold(iph
->saddr
, iph
->daddr
,
386 skb_gro_len(skb
), proto
, 0);
390 * Map a multicast IP onto multicast MAC for type ethernet.
393 static inline void ip_eth_mc_map(__be32 naddr
, char *buf
)
395 __u32 addr
=ntohl(naddr
);
407 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
408 * Leave P_Key as 0 to be filled in by driver.
411 static inline void ip_ib_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
414 unsigned char scope
= broadcast
[5] & 0xF;
416 buf
[0] = 0; /* Reserved */
417 buf
[1] = 0xff; /* Multicast QPN */
422 buf
[5] = 0x10 | scope
; /* scope from broadcast address */
423 buf
[6] = 0x40; /* IPv4 signature */
425 buf
[8] = broadcast
[8]; /* P_Key */
426 buf
[9] = broadcast
[9];
433 buf
[19] = addr
& 0xff;
435 buf
[18] = addr
& 0xff;
437 buf
[17] = addr
& 0xff;
439 buf
[16] = addr
& 0x0f;
442 static inline void ip_ipgre_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
444 if ((broadcast
[0] | broadcast
[1] | broadcast
[2] | broadcast
[3]) != 0)
445 memcpy(buf
, broadcast
, 4);
447 memcpy(buf
, &naddr
, sizeof(naddr
));
450 #if IS_ENABLED(CONFIG_IPV6)
451 #include <linux/ipv6.h>
454 static __inline__
void inet_reset_saddr(struct sock
*sk
)
456 inet_sk(sk
)->inet_rcv_saddr
= inet_sk(sk
)->inet_saddr
= 0;
457 #if IS_ENABLED(CONFIG_IPV6)
458 if (sk
->sk_family
== PF_INET6
) {
459 struct ipv6_pinfo
*np
= inet6_sk(sk
);
461 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
462 memset(&sk
->sk_v6_rcv_saddr
, 0, sizeof(sk
->sk_v6_rcv_saddr
));
469 static inline unsigned int ipv4_addr_hash(__be32 ip
)
471 return (__force
unsigned int) ip
;
474 bool ip_call_ra_chain(struct sk_buff
*skb
);
477 * Functions provided by ip_fragment.c
480 enum ip_defrag_users
{
481 IP_DEFRAG_LOCAL_DELIVER
,
482 IP_DEFRAG_CALL_RA_CHAIN
,
483 IP_DEFRAG_CONNTRACK_IN
,
484 __IP_DEFRAG_CONNTRACK_IN_END
= IP_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
485 IP_DEFRAG_CONNTRACK_OUT
,
486 __IP_DEFRAG_CONNTRACK_OUT_END
= IP_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
487 IP_DEFRAG_CONNTRACK_BRIDGE_IN
,
488 __IP_DEFRAG_CONNTRACK_BRIDGE_IN
= IP_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
496 /* Return true if the value of 'user' is between 'lower_bond'
497 * and 'upper_bond' inclusively.
499 static inline bool ip_defrag_user_in_between(u32 user
,
500 enum ip_defrag_users lower_bond
,
501 enum ip_defrag_users upper_bond
)
503 return user
>= lower_bond
&& user
<= upper_bond
;
506 int ip_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
508 struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
510 static inline struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
)
515 int ip_frag_mem(struct net
*net
);
518 * Functions provided by ip_forward.c
521 int ip_forward(struct sk_buff
*skb
);
524 * Functions provided by ip_options.c
527 void ip_options_build(struct sk_buff
*skb
, struct ip_options
*opt
,
528 __be32 daddr
, struct rtable
*rt
, int is_frag
);
530 int __ip_options_echo(struct ip_options
*dopt
, struct sk_buff
*skb
,
531 const struct ip_options
*sopt
);
532 static inline int ip_options_echo(struct ip_options
*dopt
, struct sk_buff
*skb
)
534 return __ip_options_echo(dopt
, skb
, &IPCB(skb
)->opt
);
537 void ip_options_fragment(struct sk_buff
*skb
);
538 int ip_options_compile(struct net
*net
, struct ip_options
*opt
,
539 struct sk_buff
*skb
);
540 int ip_options_get(struct net
*net
, struct ip_options_rcu
**optp
,
541 unsigned char *data
, int optlen
);
542 int ip_options_get_from_user(struct net
*net
, struct ip_options_rcu
**optp
,
543 unsigned char __user
*data
, int optlen
);
544 void ip_options_undo(struct ip_options
*opt
);
545 void ip_forward_options(struct sk_buff
*skb
);
546 int ip_options_rcv_srr(struct sk_buff
*skb
);
549 * Functions provided by ip_sockglue.c
552 void ipv4_pktinfo_prepare(const struct sock
*sk
, struct sk_buff
*skb
);
553 void ip_cmsg_recv_offset(struct msghdr
*msg
, struct sk_buff
*skb
, int offset
);
554 int ip_cmsg_send(struct sock
*sk
, struct msghdr
*msg
,
555 struct ipcm_cookie
*ipc
, bool allow_ipv6
);
556 int ip_setsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
557 unsigned int optlen
);
558 int ip_getsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
560 int compat_ip_setsockopt(struct sock
*sk
, int level
, int optname
,
561 char __user
*optval
, unsigned int optlen
);
562 int compat_ip_getsockopt(struct sock
*sk
, int level
, int optname
,
563 char __user
*optval
, int __user
*optlen
);
564 int ip_ra_control(struct sock
*sk
, unsigned char on
,
565 void (*destructor
)(struct sock
*));
567 int ip_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
, int *addr_len
);
568 void ip_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
569 u32 info
, u8
*payload
);
570 void ip_local_error(struct sock
*sk
, int err
, __be32 daddr
, __be16 dport
,
573 static inline void ip_cmsg_recv(struct msghdr
*msg
, struct sk_buff
*skb
)
575 ip_cmsg_recv_offset(msg
, skb
, 0);
578 bool icmp_global_allow(void);
579 extern int sysctl_icmp_msgs_per_sec
;
580 extern int sysctl_icmp_msgs_burst
;
582 #ifdef CONFIG_PROC_FS
583 int ip_misc_proc_init(void);