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;
61 struct ip_options_rcu
*opt
;
68 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
69 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
72 struct ip_ra_chain __rcu
*next
;
75 void (*destructor
)(struct sock
*);
76 struct sock
*saved_sk
;
81 extern struct ip_ra_chain __rcu
*ip_ra_chain
;
84 #define IP_CE 0x8000 /* Flag: "Congestion" */
85 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
86 #define IP_MF 0x2000 /* Flag: "More Fragments" */
87 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
89 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
97 int igmp_mc_init(void);
100 * Functions provided by ip.c
103 int ip_build_and_send_pkt(struct sk_buff
*skb
, const struct sock
*sk
,
104 __be32 saddr
, __be32 daddr
,
105 struct ip_options_rcu
*opt
);
106 int ip_rcv(struct sk_buff
*skb
, struct net_device
*dev
, struct packet_type
*pt
,
107 struct net_device
*orig_dev
);
108 int ip_local_deliver(struct sk_buff
*skb
);
109 int ip_mr_input(struct sk_buff
*skb
);
110 int ip_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
111 int ip_mc_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
112 int ip_do_fragment(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
,
113 int (*output
)(struct net
*, struct sock
*, struct sk_buff
*));
114 void ip_send_check(struct iphdr
*ip
);
115 int __ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
116 int ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
118 int ip_queue_xmit(struct sock
*sk
, struct sk_buff
*skb
, struct flowi
*fl
);
120 int ip_append_data(struct sock
*sk
, struct flowi4
*fl4
,
121 int getfrag(void *from
, char *to
, int offset
, int len
,
122 int odd
, struct sk_buff
*skb
),
123 void *from
, int len
, int protolen
,
124 struct ipcm_cookie
*ipc
,
127 int ip_generic_getfrag(void *from
, char *to
, int offset
, int len
, int odd
,
128 struct sk_buff
*skb
);
129 ssize_t
ip_append_page(struct sock
*sk
, struct flowi4
*fl4
, struct page
*page
,
130 int offset
, size_t size
, int flags
);
131 struct sk_buff
*__ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
132 struct sk_buff_head
*queue
,
133 struct inet_cork
*cork
);
134 int ip_send_skb(struct net
*net
, struct sk_buff
*skb
);
135 int ip_push_pending_frames(struct sock
*sk
, struct flowi4
*fl4
);
136 void ip_flush_pending_frames(struct sock
*sk
);
137 struct sk_buff
*ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
138 int getfrag(void *from
, char *to
, int offset
,
139 int len
, int odd
, struct sk_buff
*skb
),
140 void *from
, int length
, int transhdrlen
,
141 struct ipcm_cookie
*ipc
, struct rtable
**rtp
,
144 static inline struct sk_buff
*ip_finish_skb(struct sock
*sk
, struct flowi4
*fl4
)
146 return __ip_make_skb(sk
, fl4
, &sk
->sk_write_queue
, &inet_sk(sk
)->cork
.base
);
149 static inline __u8
get_rttos(struct ipcm_cookie
* ipc
, struct inet_sock
*inet
)
151 return (ipc
->tos
!= -1) ? RT_TOS(ipc
->tos
) : RT_TOS(inet
->tos
);
154 static inline __u8
get_rtconn_flags(struct ipcm_cookie
* ipc
, struct sock
* sk
)
156 return (ipc
->tos
!= -1) ? RT_CONN_FLAGS_TOS(sk
, ipc
->tos
) : RT_CONN_FLAGS(sk
);
160 int __ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
161 int ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
163 void ip4_datagram_release_cb(struct sock
*sk
);
165 struct ip_reply_arg
{
169 int csumoffset
; /* u16 offset of csum in iov[0].iov_base */
170 /* -1 if not needed */
175 #define IP_REPLY_ARG_NOSRCCHECK 1
177 static inline __u8
ip_reply_arg_flowi_flags(const struct ip_reply_arg
*arg
)
179 return (arg
->flags
& IP_REPLY_ARG_NOSRCCHECK
) ? FLOWI_FLAG_ANYSRC
: 0;
182 void ip_send_unicast_reply(struct sock
*sk
, struct sk_buff
*skb
,
183 const struct ip_options
*sopt
,
184 __be32 daddr
, __be32 saddr
,
185 const struct ip_reply_arg
*arg
,
188 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
189 #define IP_INC_STATS_BH(net, field) SNMP_INC_STATS64_BH((net)->mib.ip_statistics, field)
190 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
191 #define IP_ADD_STATS_BH(net, field, val) SNMP_ADD_STATS64_BH((net)->mib.ip_statistics, field, val)
192 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
193 #define IP_UPD_PO_STATS_BH(net, field, val) SNMP_UPD_PO_STATS64_BH((net)->mib.ip_statistics, field, val)
194 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
195 #define NET_INC_STATS_BH(net, field) SNMP_INC_STATS_BH((net)->mib.net_statistics, field)
196 #define NET_INC_STATS_USER(net, field) SNMP_INC_STATS_USER((net)->mib.net_statistics, field)
197 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
198 #define NET_ADD_STATS_BH(net, field, adnd) SNMP_ADD_STATS_BH((net)->mib.net_statistics, field, adnd)
199 #define NET_ADD_STATS_USER(net, field, adnd) SNMP_ADD_STATS_USER((net)->mib.net_statistics, field, adnd)
201 u64
snmp_get_cpu_field(void __percpu
*mib
, int cpu
, int offct
);
202 unsigned long snmp_fold_field(void __percpu
*mib
, int offt
);
203 #if BITS_PER_LONG==32
204 u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
205 size_t syncp_offset
);
206 u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t sync_off
);
208 static inline u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
211 return snmp_get_cpu_field(mib
, cpu
, offct
);
215 static inline u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t syncp_off
)
217 return snmp_fold_field(mib
, offt
);
221 void inet_get_local_port_range(struct net
*net
, int *low
, int *high
);
224 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
226 if (!net
->ipv4
.sysctl_local_reserved_ports
)
228 return test_bit(port
, net
->ipv4
.sysctl_local_reserved_ports
);
231 static inline bool sysctl_dev_name_is_allowed(const char *name
)
233 return strcmp(name
, "default") != 0 && strcmp(name
, "all") != 0;
237 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
243 /* From inetpeer.c */
244 extern int inet_peer_threshold
;
245 extern int inet_peer_minttl
;
246 extern int inet_peer_maxttl
;
248 /* From ip_input.c */
249 extern int sysctl_ip_early_demux
;
251 /* From ip_output.c */
252 extern int sysctl_ip_dynaddr
;
254 void ipfrag_init(void);
256 void ip_static_sysctl_init(void);
258 #define IP4_REPLY_MARK(net, mark) \
259 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
261 static inline bool ip_is_fragment(const struct iphdr
*iph
)
263 return (iph
->frag_off
& htons(IP_MF
| IP_OFFSET
)) != 0;
269 /* The function in 2.2 was invalid, producing wrong result for
270 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
272 int ip_decrease_ttl(struct iphdr
*iph
)
274 u32 check
= (__force u32
)iph
->check
;
275 check
+= (__force u32
)htons(0x0100);
276 iph
->check
= (__force __sum16
)(check
+ (check
>=0xFFFF));
281 int ip_dont_fragment(const struct sock
*sk
, const struct dst_entry
*dst
)
283 u8 pmtudisc
= READ_ONCE(inet_sk(sk
)->pmtudisc
);
285 return pmtudisc
== IP_PMTUDISC_DO
||
286 (pmtudisc
== IP_PMTUDISC_WANT
&&
287 !(dst_metric_locked(dst
, RTAX_MTU
)));
290 static inline bool ip_sk_accept_pmtu(const struct sock
*sk
)
292 return inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_INTERFACE
&&
293 inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_OMIT
;
296 static inline bool ip_sk_use_pmtu(const struct sock
*sk
)
298 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_PROBE
;
301 static inline bool ip_sk_ignore_df(const struct sock
*sk
)
303 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_DO
||
304 inet_sk(sk
)->pmtudisc
== IP_PMTUDISC_OMIT
;
307 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry
*dst
,
310 struct net
*net
= dev_net(dst
->dev
);
312 if (net
->ipv4
.sysctl_ip_fwd_use_pmtu
||
313 dst_metric_locked(dst
, RTAX_MTU
) ||
317 return min(READ_ONCE(dst
->dev
->mtu
), IP_MAX_MTU
);
320 static inline unsigned int ip_skb_dst_mtu(const struct sk_buff
*skb
)
322 struct sock
*sk
= skb
->sk
;
324 if (!sk
|| !sk_fullsock(sk
) || ip_sk_use_pmtu(sk
)) {
325 bool forwarding
= IPCB(skb
)->flags
& IPSKB_FORWARDED
;
327 return ip_dst_mtu_maybe_forward(skb_dst(skb
), forwarding
);
330 return min(READ_ONCE(skb_dst(skb
)->dev
->mtu
), IP_MAX_MTU
);
333 u32
ip_idents_reserve(u32 hash
, int segs
);
334 void __ip_select_ident(struct net
*net
, struct iphdr
*iph
, int segs
);
336 static inline void ip_select_ident_segs(struct net
*net
, struct sk_buff
*skb
,
337 struct sock
*sk
, int segs
)
339 struct iphdr
*iph
= ip_hdr(skb
);
341 if ((iph
->frag_off
& htons(IP_DF
)) && !skb
->ignore_df
) {
342 /* This is only to work around buggy Windows95/2000
343 * VJ compression implementations. If the ID field
344 * does not change, they drop every other packet in
345 * a TCP stream using header compression.
347 if (sk
&& inet_sk(sk
)->inet_daddr
) {
348 iph
->id
= htons(inet_sk(sk
)->inet_id
);
349 inet_sk(sk
)->inet_id
+= segs
;
354 __ip_select_ident(net
, iph
, segs
);
358 static inline void ip_select_ident(struct net
*net
, struct sk_buff
*skb
,
361 ip_select_ident_segs(net
, skb
, sk
, 1);
364 static inline __wsum
inet_compute_pseudo(struct sk_buff
*skb
, int proto
)
366 return csum_tcpudp_nofold(ip_hdr(skb
)->saddr
, ip_hdr(skb
)->daddr
,
370 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
371 * Equivalent to : flow->v4addrs.src = iph->saddr;
372 * flow->v4addrs.dst = iph->daddr;
374 static inline void iph_to_flow_copy_v4addrs(struct flow_keys
*flow
,
375 const struct iphdr
*iph
)
377 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v4addrs
.dst
) !=
378 offsetof(typeof(flow
->addrs
), v4addrs
.src
) +
379 sizeof(flow
->addrs
.v4addrs
.src
));
380 memcpy(&flow
->addrs
.v4addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v4addrs
));
381 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV4_ADDRS
;
384 static inline __wsum
inet_gro_compute_pseudo(struct sk_buff
*skb
, int proto
)
386 const struct iphdr
*iph
= skb_gro_network_header(skb
);
388 return csum_tcpudp_nofold(iph
->saddr
, iph
->daddr
,
389 skb_gro_len(skb
), proto
, 0);
393 * Map a multicast IP onto multicast MAC for type ethernet.
396 static inline void ip_eth_mc_map(__be32 naddr
, char *buf
)
398 __u32 addr
=ntohl(naddr
);
410 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
411 * Leave P_Key as 0 to be filled in by driver.
414 static inline void ip_ib_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
417 unsigned char scope
= broadcast
[5] & 0xF;
419 buf
[0] = 0; /* Reserved */
420 buf
[1] = 0xff; /* Multicast QPN */
425 buf
[5] = 0x10 | scope
; /* scope from broadcast address */
426 buf
[6] = 0x40; /* IPv4 signature */
428 buf
[8] = broadcast
[8]; /* P_Key */
429 buf
[9] = broadcast
[9];
436 buf
[19] = addr
& 0xff;
438 buf
[18] = addr
& 0xff;
440 buf
[17] = addr
& 0xff;
442 buf
[16] = addr
& 0x0f;
445 static inline void ip_ipgre_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
447 if ((broadcast
[0] | broadcast
[1] | broadcast
[2] | broadcast
[3]) != 0)
448 memcpy(buf
, broadcast
, 4);
450 memcpy(buf
, &naddr
, sizeof(naddr
));
453 #if IS_ENABLED(CONFIG_IPV6)
454 #include <linux/ipv6.h>
457 static __inline__
void inet_reset_saddr(struct sock
*sk
)
459 inet_sk(sk
)->inet_rcv_saddr
= inet_sk(sk
)->inet_saddr
= 0;
460 #if IS_ENABLED(CONFIG_IPV6)
461 if (sk
->sk_family
== PF_INET6
) {
462 struct ipv6_pinfo
*np
= inet6_sk(sk
);
464 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
465 memset(&sk
->sk_v6_rcv_saddr
, 0, sizeof(sk
->sk_v6_rcv_saddr
));
472 static inline unsigned int ipv4_addr_hash(__be32 ip
)
474 return (__force
unsigned int) ip
;
477 bool ip_call_ra_chain(struct sk_buff
*skb
);
480 * Functions provided by ip_fragment.c
483 enum ip_defrag_users
{
484 IP_DEFRAG_LOCAL_DELIVER
,
485 IP_DEFRAG_CALL_RA_CHAIN
,
486 IP_DEFRAG_CONNTRACK_IN
,
487 __IP_DEFRAG_CONNTRACK_IN_END
= IP_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
488 IP_DEFRAG_CONNTRACK_OUT
,
489 __IP_DEFRAG_CONNTRACK_OUT_END
= IP_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
490 IP_DEFRAG_CONNTRACK_BRIDGE_IN
,
491 __IP_DEFRAG_CONNTRACK_BRIDGE_IN
= IP_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
499 /* Return true if the value of 'user' is between 'lower_bond'
500 * and 'upper_bond' inclusively.
502 static inline bool ip_defrag_user_in_between(u32 user
,
503 enum ip_defrag_users lower_bond
,
504 enum ip_defrag_users upper_bond
)
506 return user
>= lower_bond
&& user
<= upper_bond
;
509 int ip_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
511 struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
513 static inline struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
)
518 int ip_frag_mem(struct net
*net
);
521 * Functions provided by ip_forward.c
524 int ip_forward(struct sk_buff
*skb
);
527 * Functions provided by ip_options.c
530 void ip_options_build(struct sk_buff
*skb
, struct ip_options
*opt
,
531 __be32 daddr
, struct rtable
*rt
, int is_frag
);
533 int __ip_options_echo(struct ip_options
*dopt
, struct sk_buff
*skb
,
534 const struct ip_options
*sopt
);
535 static inline int ip_options_echo(struct ip_options
*dopt
, struct sk_buff
*skb
)
537 return __ip_options_echo(dopt
, skb
, &IPCB(skb
)->opt
);
540 void ip_options_fragment(struct sk_buff
*skb
);
541 int ip_options_compile(struct net
*net
, struct ip_options
*opt
,
542 struct sk_buff
*skb
);
543 int ip_options_get(struct net
*net
, struct ip_options_rcu
**optp
,
544 unsigned char *data
, int optlen
);
545 int ip_options_get_from_user(struct net
*net
, struct ip_options_rcu
**optp
,
546 unsigned char __user
*data
, int optlen
);
547 void ip_options_undo(struct ip_options
*opt
);
548 void ip_forward_options(struct sk_buff
*skb
);
549 int ip_options_rcv_srr(struct sk_buff
*skb
);
552 * Functions provided by ip_sockglue.c
555 void ipv4_pktinfo_prepare(const struct sock
*sk
, struct sk_buff
*skb
);
556 void ip_cmsg_recv_offset(struct msghdr
*msg
, struct sk_buff
*skb
, int tlen
, int offset
);
557 int ip_cmsg_send(struct net
*net
, struct msghdr
*msg
,
558 struct ipcm_cookie
*ipc
, bool allow_ipv6
);
559 int ip_setsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
560 unsigned int optlen
);
561 int ip_getsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
563 int compat_ip_setsockopt(struct sock
*sk
, int level
, int optname
,
564 char __user
*optval
, unsigned int optlen
);
565 int compat_ip_getsockopt(struct sock
*sk
, int level
, int optname
,
566 char __user
*optval
, int __user
*optlen
);
567 int ip_ra_control(struct sock
*sk
, unsigned char on
,
568 void (*destructor
)(struct sock
*));
570 int ip_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
, int *addr_len
);
571 void ip_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
572 u32 info
, u8
*payload
);
573 void ip_local_error(struct sock
*sk
, int err
, __be32 daddr
, __be16 dport
,
576 static inline void ip_cmsg_recv(struct msghdr
*msg
, struct sk_buff
*skb
)
578 ip_cmsg_recv_offset(msg
, skb
, 0, 0);
581 bool icmp_global_allow(void);
582 extern int sysctl_icmp_msgs_per_sec
;
583 extern int sysctl_icmp_msgs_burst
;
585 #ifdef CONFIG_PROC_FS
586 int ip_misc_proc_init(void);