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 */
37 #define IPV4_MIN_MTU 68 /* RFC 791 */
41 struct inet_skb_parm
{
43 struct ip_options opt
; /* Compiled IP options */
46 #define IPSKB_FORWARDED BIT(0)
47 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
48 #define IPSKB_XFRM_TRANSFORMED BIT(2)
49 #define IPSKB_FRAG_COMPLETE BIT(3)
50 #define IPSKB_REROUTED BIT(4)
51 #define IPSKB_DOREDIRECT BIT(5)
52 #define IPSKB_FRAG_PMTU BIT(6)
53 #define IPSKB_L3SLAVE BIT(7)
58 static inline bool ipv4_l3mdev_skb(u16 flags
)
60 return !!(flags
& IPSKB_L3SLAVE
);
63 static inline unsigned int ip_hdrlen(const struct sk_buff
*skb
)
65 return ip_hdr(skb
)->ihl
* 4;
69 struct sockcm_cookie sockc
;
72 struct ip_options_rcu
*opt
;
79 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
80 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
82 /* return enslaved device index if relevant */
83 static inline int inet_sdif(struct sk_buff
*skb
)
85 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
86 if (skb
&& ipv4_l3mdev_skb(IPCB(skb
)->flags
))
87 return IPCB(skb
)->iif
;
93 struct ip_ra_chain __rcu
*next
;
96 void (*destructor
)(struct sock
*);
97 struct sock
*saved_sk
;
102 extern struct ip_ra_chain __rcu
*ip_ra_chain
;
105 #define IP_CE 0x8000 /* Flag: "Congestion" */
106 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
107 #define IP_MF 0x2000 /* Flag: "More Fragments" */
108 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
110 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
118 int igmp_mc_init(void);
121 * Functions provided by ip.c
124 int ip_build_and_send_pkt(struct sk_buff
*skb
, const struct sock
*sk
,
125 __be32 saddr
, __be32 daddr
,
126 struct ip_options_rcu
*opt
);
127 int ip_rcv(struct sk_buff
*skb
, struct net_device
*dev
, struct packet_type
*pt
,
128 struct net_device
*orig_dev
);
129 int ip_local_deliver(struct sk_buff
*skb
);
130 int ip_mr_input(struct sk_buff
*skb
);
131 int ip_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
132 int ip_mc_output(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
133 int ip_do_fragment(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
,
134 int (*output
)(struct net
*, struct sock
*, struct sk_buff
*));
135 void ip_send_check(struct iphdr
*ip
);
136 int __ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
137 int ip_local_out(struct net
*net
, struct sock
*sk
, struct sk_buff
*skb
);
139 int ip_queue_xmit(struct sock
*sk
, struct sk_buff
*skb
, struct flowi
*fl
);
141 int ip_append_data(struct sock
*sk
, struct flowi4
*fl4
,
142 int getfrag(void *from
, char *to
, int offset
, int len
,
143 int odd
, struct sk_buff
*skb
),
144 void *from
, int len
, int protolen
,
145 struct ipcm_cookie
*ipc
,
148 int ip_generic_getfrag(void *from
, char *to
, int offset
, int len
, int odd
,
149 struct sk_buff
*skb
);
150 ssize_t
ip_append_page(struct sock
*sk
, struct flowi4
*fl4
, struct page
*page
,
151 int offset
, size_t size
, int flags
);
152 struct sk_buff
*__ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
153 struct sk_buff_head
*queue
,
154 struct inet_cork
*cork
);
155 int ip_send_skb(struct net
*net
, struct sk_buff
*skb
);
156 int ip_push_pending_frames(struct sock
*sk
, struct flowi4
*fl4
);
157 void ip_flush_pending_frames(struct sock
*sk
);
158 struct sk_buff
*ip_make_skb(struct sock
*sk
, struct flowi4
*fl4
,
159 int getfrag(void *from
, char *to
, int offset
,
160 int len
, int odd
, struct sk_buff
*skb
),
161 void *from
, int length
, int transhdrlen
,
162 struct ipcm_cookie
*ipc
, struct rtable
**rtp
,
165 static inline struct sk_buff
*ip_finish_skb(struct sock
*sk
, struct flowi4
*fl4
)
167 return __ip_make_skb(sk
, fl4
, &sk
->sk_write_queue
, &inet_sk(sk
)->cork
.base
);
170 static inline __u8
get_rttos(struct ipcm_cookie
* ipc
, struct inet_sock
*inet
)
172 return (ipc
->tos
!= -1) ? RT_TOS(ipc
->tos
) : RT_TOS(inet
->tos
);
175 static inline __u8
get_rtconn_flags(struct ipcm_cookie
* ipc
, struct sock
* sk
)
177 return (ipc
->tos
!= -1) ? RT_CONN_FLAGS_TOS(sk
, ipc
->tos
) : RT_CONN_FLAGS(sk
);
181 int __ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
182 int ip4_datagram_connect(struct sock
*sk
, struct sockaddr
*uaddr
, int addr_len
);
184 void ip4_datagram_release_cb(struct sock
*sk
);
186 struct ip_reply_arg
{
190 int csumoffset
; /* u16 offset of csum in iov[0].iov_base */
191 /* -1 if not needed */
197 #define IP_REPLY_ARG_NOSRCCHECK 1
199 static inline __u8
ip_reply_arg_flowi_flags(const struct ip_reply_arg
*arg
)
201 return (arg
->flags
& IP_REPLY_ARG_NOSRCCHECK
) ? FLOWI_FLAG_ANYSRC
: 0;
204 void ip_send_unicast_reply(struct sock
*sk
, struct sk_buff
*skb
,
205 const struct ip_options
*sopt
,
206 __be32 daddr
, __be32 saddr
,
207 const struct ip_reply_arg
*arg
,
210 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
211 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
212 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
213 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_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 __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
216 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
217 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
218 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
219 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
221 u64
snmp_get_cpu_field(void __percpu
*mib
, int cpu
, int offct
);
222 unsigned long snmp_fold_field(void __percpu
*mib
, int offt
);
223 #if BITS_PER_LONG==32
224 u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
225 size_t syncp_offset
);
226 u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t sync_off
);
228 static inline u64
snmp_get_cpu_field64(void __percpu
*mib
, int cpu
, int offct
,
231 return snmp_get_cpu_field(mib
, cpu
, offct
);
235 static inline u64
snmp_fold_field64(void __percpu
*mib
, int offt
, size_t syncp_off
)
237 return snmp_fold_field(mib
, offt
);
241 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
244 for_each_possible_cpu(c) { \
245 for (i = 0; stats_list[i].name; i++) \
246 buff64[i] += snmp_get_cpu_field64( \
248 c, stats_list[i].entry, \
253 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
256 for_each_possible_cpu(c) { \
257 for (i = 0; stats_list[i].name; i++) \
258 buff[i] += snmp_get_cpu_field( \
260 c, stats_list[i].entry); \
264 void inet_get_local_port_range(struct net
*net
, int *low
, int *high
);
267 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
269 if (!net
->ipv4
.sysctl_local_reserved_ports
)
271 return test_bit(port
, net
->ipv4
.sysctl_local_reserved_ports
);
274 static inline bool sysctl_dev_name_is_allowed(const char *name
)
276 return strcmp(name
, "default") != 0 && strcmp(name
, "all") != 0;
279 static inline int inet_prot_sock(struct net
*net
)
281 return net
->ipv4
.sysctl_ip_prot_sock
;
285 static inline int inet_is_local_reserved_port(struct net
*net
, int port
)
290 static inline int inet_prot_sock(struct net
*net
)
296 __be32
inet_current_timestamp(void);
298 /* From inetpeer.c */
299 extern int inet_peer_threshold
;
300 extern int inet_peer_minttl
;
301 extern int inet_peer_maxttl
;
303 void ipfrag_init(void);
305 void ip_static_sysctl_init(void);
307 #define IP4_REPLY_MARK(net, mark) \
308 ((net)->ipv4.sysctl_fwmark_reflect ? (mark) : 0)
310 static inline bool ip_is_fragment(const struct iphdr
*iph
)
312 return (iph
->frag_off
& htons(IP_MF
| IP_OFFSET
)) != 0;
318 /* The function in 2.2 was invalid, producing wrong result for
319 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
321 int ip_decrease_ttl(struct iphdr
*iph
)
323 u32 check
= (__force u32
)iph
->check
;
324 check
+= (__force u32
)htons(0x0100);
325 iph
->check
= (__force __sum16
)(check
+ (check
>=0xFFFF));
330 int ip_dont_fragment(const struct sock
*sk
, const struct dst_entry
*dst
)
332 u8 pmtudisc
= READ_ONCE(inet_sk(sk
)->pmtudisc
);
334 return pmtudisc
== IP_PMTUDISC_DO
||
335 (pmtudisc
== IP_PMTUDISC_WANT
&&
336 !(dst_metric_locked(dst
, RTAX_MTU
)));
339 static inline bool ip_sk_accept_pmtu(const struct sock
*sk
)
341 return inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_INTERFACE
&&
342 inet_sk(sk
)->pmtudisc
!= IP_PMTUDISC_OMIT
;
345 static inline bool ip_sk_use_pmtu(const struct sock
*sk
)
347 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_PROBE
;
350 static inline bool ip_sk_ignore_df(const struct sock
*sk
)
352 return inet_sk(sk
)->pmtudisc
< IP_PMTUDISC_DO
||
353 inet_sk(sk
)->pmtudisc
== IP_PMTUDISC_OMIT
;
356 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry
*dst
,
359 struct net
*net
= dev_net(dst
->dev
);
361 if (net
->ipv4
.sysctl_ip_fwd_use_pmtu
||
362 dst_metric_locked(dst
, RTAX_MTU
) ||
366 return min(READ_ONCE(dst
->dev
->mtu
), IP_MAX_MTU
);
369 static inline unsigned int ip_skb_dst_mtu(struct sock
*sk
,
370 const struct sk_buff
*skb
)
372 if (!sk
|| !sk_fullsock(sk
) || ip_sk_use_pmtu(sk
)) {
373 bool forwarding
= IPCB(skb
)->flags
& IPSKB_FORWARDED
;
375 return ip_dst_mtu_maybe_forward(skb_dst(skb
), forwarding
);
378 return min(READ_ONCE(skb_dst(skb
)->dev
->mtu
), IP_MAX_MTU
);
381 u32
ip_idents_reserve(u32 hash
, int segs
);
382 void __ip_select_ident(struct net
*net
, struct iphdr
*iph
, int segs
);
384 static inline void ip_select_ident_segs(struct net
*net
, struct sk_buff
*skb
,
385 struct sock
*sk
, int segs
)
387 struct iphdr
*iph
= ip_hdr(skb
);
389 if ((iph
->frag_off
& htons(IP_DF
)) && !skb
->ignore_df
) {
390 /* This is only to work around buggy Windows95/2000
391 * VJ compression implementations. If the ID field
392 * does not change, they drop every other packet in
393 * a TCP stream using header compression.
395 if (sk
&& inet_sk(sk
)->inet_daddr
) {
396 iph
->id
= htons(inet_sk(sk
)->inet_id
);
397 inet_sk(sk
)->inet_id
+= segs
;
402 __ip_select_ident(net
, iph
, segs
);
406 static inline void ip_select_ident(struct net
*net
, struct sk_buff
*skb
,
409 ip_select_ident_segs(net
, skb
, sk
, 1);
412 static inline __wsum
inet_compute_pseudo(struct sk_buff
*skb
, int proto
)
414 return csum_tcpudp_nofold(ip_hdr(skb
)->saddr
, ip_hdr(skb
)->daddr
,
418 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
419 * Equivalent to : flow->v4addrs.src = iph->saddr;
420 * flow->v4addrs.dst = iph->daddr;
422 static inline void iph_to_flow_copy_v4addrs(struct flow_keys
*flow
,
423 const struct iphdr
*iph
)
425 BUILD_BUG_ON(offsetof(typeof(flow
->addrs
), v4addrs
.dst
) !=
426 offsetof(typeof(flow
->addrs
), v4addrs
.src
) +
427 sizeof(flow
->addrs
.v4addrs
.src
));
428 memcpy(&flow
->addrs
.v4addrs
, &iph
->saddr
, sizeof(flow
->addrs
.v4addrs
));
429 flow
->control
.addr_type
= FLOW_DISSECTOR_KEY_IPV4_ADDRS
;
432 static inline __wsum
inet_gro_compute_pseudo(struct sk_buff
*skb
, int proto
)
434 const struct iphdr
*iph
= skb_gro_network_header(skb
);
436 return csum_tcpudp_nofold(iph
->saddr
, iph
->daddr
,
437 skb_gro_len(skb
), proto
, 0);
441 * Map a multicast IP onto multicast MAC for type ethernet.
444 static inline void ip_eth_mc_map(__be32 naddr
, char *buf
)
446 __u32 addr
=ntohl(naddr
);
458 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
459 * Leave P_Key as 0 to be filled in by driver.
462 static inline void ip_ib_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
465 unsigned char scope
= broadcast
[5] & 0xF;
467 buf
[0] = 0; /* Reserved */
468 buf
[1] = 0xff; /* Multicast QPN */
473 buf
[5] = 0x10 | scope
; /* scope from broadcast address */
474 buf
[6] = 0x40; /* IPv4 signature */
476 buf
[8] = broadcast
[8]; /* P_Key */
477 buf
[9] = broadcast
[9];
484 buf
[19] = addr
& 0xff;
486 buf
[18] = addr
& 0xff;
488 buf
[17] = addr
& 0xff;
490 buf
[16] = addr
& 0x0f;
493 static inline void ip_ipgre_mc_map(__be32 naddr
, const unsigned char *broadcast
, char *buf
)
495 if ((broadcast
[0] | broadcast
[1] | broadcast
[2] | broadcast
[3]) != 0)
496 memcpy(buf
, broadcast
, 4);
498 memcpy(buf
, &naddr
, sizeof(naddr
));
501 #if IS_ENABLED(CONFIG_IPV6)
502 #include <linux/ipv6.h>
505 static __inline__
void inet_reset_saddr(struct sock
*sk
)
507 inet_sk(sk
)->inet_rcv_saddr
= inet_sk(sk
)->inet_saddr
= 0;
508 #if IS_ENABLED(CONFIG_IPV6)
509 if (sk
->sk_family
== PF_INET6
) {
510 struct ipv6_pinfo
*np
= inet6_sk(sk
);
512 memset(&np
->saddr
, 0, sizeof(np
->saddr
));
513 memset(&sk
->sk_v6_rcv_saddr
, 0, sizeof(sk
->sk_v6_rcv_saddr
));
520 static inline unsigned int ipv4_addr_hash(__be32 ip
)
522 return (__force
unsigned int) ip
;
525 bool ip_call_ra_chain(struct sk_buff
*skb
);
528 * Functions provided by ip_fragment.c
531 enum ip_defrag_users
{
532 IP_DEFRAG_LOCAL_DELIVER
,
533 IP_DEFRAG_CALL_RA_CHAIN
,
534 IP_DEFRAG_CONNTRACK_IN
,
535 __IP_DEFRAG_CONNTRACK_IN_END
= IP_DEFRAG_CONNTRACK_IN
+ USHRT_MAX
,
536 IP_DEFRAG_CONNTRACK_OUT
,
537 __IP_DEFRAG_CONNTRACK_OUT_END
= IP_DEFRAG_CONNTRACK_OUT
+ USHRT_MAX
,
538 IP_DEFRAG_CONNTRACK_BRIDGE_IN
,
539 __IP_DEFRAG_CONNTRACK_BRIDGE_IN
= IP_DEFRAG_CONNTRACK_BRIDGE_IN
+ USHRT_MAX
,
547 /* Return true if the value of 'user' is between 'lower_bond'
548 * and 'upper_bond' inclusively.
550 static inline bool ip_defrag_user_in_between(u32 user
,
551 enum ip_defrag_users lower_bond
,
552 enum ip_defrag_users upper_bond
)
554 return user
>= lower_bond
&& user
<= upper_bond
;
557 int ip_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
559 struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
);
561 static inline struct sk_buff
*ip_check_defrag(struct net
*net
, struct sk_buff
*skb
, u32 user
)
566 int ip_frag_mem(struct net
*net
);
569 * Functions provided by ip_forward.c
572 int ip_forward(struct sk_buff
*skb
);
575 * Functions provided by ip_options.c
578 void ip_options_build(struct sk_buff
*skb
, struct ip_options
*opt
,
579 __be32 daddr
, struct rtable
*rt
, int is_frag
);
581 int __ip_options_echo(struct net
*net
, struct ip_options
*dopt
,
582 struct sk_buff
*skb
, const struct ip_options
*sopt
);
583 static inline int ip_options_echo(struct net
*net
, struct ip_options
*dopt
,
586 return __ip_options_echo(net
, dopt
, skb
, &IPCB(skb
)->opt
);
589 void ip_options_fragment(struct sk_buff
*skb
);
590 int ip_options_compile(struct net
*net
, struct ip_options
*opt
,
591 struct sk_buff
*skb
);
592 int ip_options_get(struct net
*net
, struct ip_options_rcu
**optp
,
593 unsigned char *data
, int optlen
);
594 int ip_options_get_from_user(struct net
*net
, struct ip_options_rcu
**optp
,
595 unsigned char __user
*data
, int optlen
);
596 void ip_options_undo(struct ip_options
*opt
);
597 void ip_forward_options(struct sk_buff
*skb
);
598 int ip_options_rcv_srr(struct sk_buff
*skb
);
601 * Functions provided by ip_sockglue.c
604 void ipv4_pktinfo_prepare(const struct sock
*sk
, struct sk_buff
*skb
);
605 void ip_cmsg_recv_offset(struct msghdr
*msg
, struct sock
*sk
,
606 struct sk_buff
*skb
, int tlen
, int offset
);
607 int ip_cmsg_send(struct sock
*sk
, struct msghdr
*msg
,
608 struct ipcm_cookie
*ipc
, bool allow_ipv6
);
609 int ip_setsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
610 unsigned int optlen
);
611 int ip_getsockopt(struct sock
*sk
, int level
, int optname
, char __user
*optval
,
613 int compat_ip_setsockopt(struct sock
*sk
, int level
, int optname
,
614 char __user
*optval
, unsigned int optlen
);
615 int compat_ip_getsockopt(struct sock
*sk
, int level
, int optname
,
616 char __user
*optval
, int __user
*optlen
);
617 int ip_ra_control(struct sock
*sk
, unsigned char on
,
618 void (*destructor
)(struct sock
*));
620 int ip_recv_error(struct sock
*sk
, struct msghdr
*msg
, int len
, int *addr_len
);
621 void ip_icmp_error(struct sock
*sk
, struct sk_buff
*skb
, int err
, __be16 port
,
622 u32 info
, u8
*payload
);
623 void ip_local_error(struct sock
*sk
, int err
, __be32 daddr
, __be16 dport
,
626 static inline void ip_cmsg_recv(struct msghdr
*msg
, struct sk_buff
*skb
)
628 ip_cmsg_recv_offset(msg
, skb
->sk
, skb
, 0, 0);
631 bool icmp_global_allow(void);
632 extern int sysctl_icmp_msgs_per_sec
;
633 extern int sysctl_icmp_msgs_burst
;
635 #ifdef CONFIG_PROC_FS
636 int ip_misc_proc_init(void);