1 /* SPDX-License-Identifier: GPL-2.0-or-later */
2 /*
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
6 *
7 * Definitions for the IP module.
8 *
9 * Version: @(#)ip.h 1.0.2 05/07/93
10 *
11 * Authors: Ross Biro
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Alan Cox, <gw4pts@gw4pts.ampr.org>
14 *
15 * Changes:
16 * Mike McLagan : Routing by source
17 */
18 #ifndef _IP_H
19 #define _IP_H
20
21 #include <linux/types.h>
22 #include <linux/ip.h>
23 #include <linux/in.h>
24 #include <linux/skbuff.h>
25 #include <linux/jhash.h>
26 #include <linux/sockptr.h>
27 #include <linux/static_key.h>
28 #include <linux/android_kabi.h>
29
30 #include <net/inet_sock.h>
31 #include <net/route.h>
32 #include <net/snmp.h>
33 #include <net/flow.h>
34 #include <net/flow_dissector.h>
35 #include <net/netns/hash.h>
36 #include <net/lwtunnel.h>
37 #include <net/inet_dscp.h>
38
39 #define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */
40 #define IPV4_MIN_MTU 68 /* RFC 791 */
41
42 extern unsigned int sysctl_fib_sync_mem;
43 extern unsigned int sysctl_fib_sync_mem_min;
44 extern unsigned int sysctl_fib_sync_mem_max;
45
46 struct sock;
47
48 struct inet_skb_parm {
49 int iif;
50 struct ip_options opt; /* Compiled IP options */
51 u16 flags;
52
53 #define IPSKB_FORWARDED BIT(0)
54 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
55 #define IPSKB_XFRM_TRANSFORMED BIT(2)
56 #define IPSKB_FRAG_COMPLETE BIT(3)
57 #define IPSKB_REROUTED BIT(4)
58 #define IPSKB_DOREDIRECT BIT(5)
59 #define IPSKB_FRAG_PMTU BIT(6)
60 #define IPSKB_L3SLAVE BIT(7)
61 #define IPSKB_NOPOLICY BIT(8)
62 #define IPSKB_MULTIPATH BIT(9)
63
64 u16 frag_max_size;
65 };
66
ipv4_l3mdev_skb(u16 flags)67 static inline bool ipv4_l3mdev_skb(u16 flags)
68 {
69 return !!(flags & IPSKB_L3SLAVE);
70 }
71
ip_hdrlen(const struct sk_buff * skb)72 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
73 {
74 return ip_hdr(skb)->ihl * 4;
75 }
76
77 struct ipcm_cookie {
78 struct sockcm_cookie sockc;
79 __be32 addr;
80 int oif;
81 struct ip_options_rcu *opt;
82 __u8 protocol;
83 __u8 ttl;
84 __s16 tos;
85 char priority;
86 __u16 gso_size;
87 ANDROID_KABI_RESERVE(1);
88 };
89
ipcm_init(struct ipcm_cookie * ipcm)90 static inline void ipcm_init(struct ipcm_cookie *ipcm)
91 {
92 *ipcm = (struct ipcm_cookie) { .tos = -1 };
93 }
94
ipcm_init_sk(struct ipcm_cookie * ipcm,const struct inet_sock * inet)95 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
96 const struct inet_sock *inet)
97 {
98 ipcm_init(ipcm);
99
100 ipcm->sockc.mark = READ_ONCE(inet->sk.sk_mark);
101 ipcm->sockc.tsflags = READ_ONCE(inet->sk.sk_tsflags);
102 ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if);
103 ipcm->addr = inet->inet_saddr;
104 ipcm->protocol = inet->inet_num;
105 }
106
107 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
108 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
109
110 /* return enslaved device index if relevant */
inet_sdif(const struct sk_buff * skb)111 static inline int inet_sdif(const struct sk_buff *skb)
112 {
113 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
114 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
115 return IPCB(skb)->iif;
116 #endif
117 return 0;
118 }
119
120 /* Special input handler for packets caught by router alert option.
121 They are selected only by protocol field, and then processed likely
122 local ones; but only if someone wants them! Otherwise, router
123 not running rsvpd will kill RSVP.
124
125 It is user level problem, what it will make with them.
126 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
127 but receiver should be enough clever f.e. to forward mtrace requests,
128 sent to multicast group to reach destination designated router.
129 */
130
131 struct ip_ra_chain {
132 struct ip_ra_chain __rcu *next;
133 struct sock *sk;
134 union {
135 void (*destructor)(struct sock *);
136 struct sock *saved_sk;
137 };
138 struct rcu_head rcu;
139 };
140
141 /* IP flags. */
142 #define IP_CE 0x8000 /* Flag: "Congestion" */
143 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
144 #define IP_MF 0x2000 /* Flag: "More Fragments" */
145 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
146
147 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
148
149 struct msghdr;
150 struct net_device;
151 struct packet_type;
152 struct rtable;
153 struct sockaddr;
154
155 int igmp_mc_init(void);
156
157 /*
158 * Functions provided by ip.c
159 */
160
161 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
162 __be32 saddr, __be32 daddr,
163 struct ip_options_rcu *opt, u8 tos);
164 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
165 struct net_device *orig_dev);
166 void ip_list_rcv(struct list_head *head, struct packet_type *pt,
167 struct net_device *orig_dev);
168 int ip_local_deliver(struct sk_buff *skb);
169 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
170 int ip_mr_input(struct sk_buff *skb);
171 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
172 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
173 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
174 int (*output)(struct net *, struct sock *, struct sk_buff *));
175
176 struct ip_fraglist_iter {
177 struct sk_buff *frag;
178 struct iphdr *iph;
179 int offset;
180 unsigned int hlen;
181 };
182
183 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
184 unsigned int hlen, struct ip_fraglist_iter *iter);
185 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
186
ip_fraglist_next(struct ip_fraglist_iter * iter)187 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
188 {
189 struct sk_buff *skb = iter->frag;
190
191 iter->frag = skb->next;
192 skb_mark_not_on_list(skb);
193
194 return skb;
195 }
196
197 struct ip_frag_state {
198 bool DF;
199 unsigned int hlen;
200 unsigned int ll_rs;
201 unsigned int mtu;
202 unsigned int left;
203 int offset;
204 int ptr;
205 __be16 not_last_frag;
206 };
207
208 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
209 unsigned int mtu, bool DF, struct ip_frag_state *state);
210 struct sk_buff *ip_frag_next(struct sk_buff *skb,
211 struct ip_frag_state *state);
212
213 void ip_send_check(struct iphdr *ip);
214 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
215 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
216
217 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
218 __u8 tos);
219 void ip_init(void);
220 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
221 int getfrag(void *from, char *to, int offset, int len,
222 int odd, struct sk_buff *skb),
223 void *from, int len, int protolen,
224 struct ipcm_cookie *ipc,
225 struct rtable **rt,
226 unsigned int flags);
227 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
228 struct sk_buff *skb);
229 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
230 struct sk_buff_head *queue,
231 struct inet_cork *cork);
232 int ip_send_skb(struct net *net, struct sk_buff *skb);
233 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
234 void ip_flush_pending_frames(struct sock *sk);
235 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
236 int getfrag(void *from, char *to, int offset,
237 int len, int odd, struct sk_buff *skb),
238 void *from, int length, int transhdrlen,
239 struct ipcm_cookie *ipc, struct rtable **rtp,
240 struct inet_cork *cork, unsigned int flags);
241
242 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
243
ip_finish_skb(struct sock * sk,struct flowi4 * fl4)244 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
245 {
246 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
247 }
248
249 /* Get the route scope that should be used when sending a packet. */
ip_sendmsg_scope(const struct inet_sock * inet,const struct ipcm_cookie * ipc,const struct msghdr * msg)250 static inline u8 ip_sendmsg_scope(const struct inet_sock *inet,
251 const struct ipcm_cookie *ipc,
252 const struct msghdr *msg)
253 {
254 if (sock_flag(&inet->sk, SOCK_LOCALROUTE) ||
255 msg->msg_flags & MSG_DONTROUTE ||
256 (ipc->opt && ipc->opt->opt.is_strictroute))
257 return RT_SCOPE_LINK;
258
259 return RT_SCOPE_UNIVERSE;
260 }
261
get_rttos(struct ipcm_cookie * ipc,struct inet_sock * inet)262 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
263 {
264 u8 dsfield = ipc->tos != -1 ? ipc->tos : READ_ONCE(inet->tos);
265
266 return dsfield & INET_DSCP_MASK;
267 }
268
269 /* datagram.c */
270 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
271 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
272
273 void ip4_datagram_release_cb(struct sock *sk);
274
275 struct ip_reply_arg {
276 struct kvec iov[1];
277 int flags;
278 __wsum csum;
279 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
280 /* -1 if not needed */
281 int bound_dev_if;
282 u8 tos;
283 kuid_t uid;
284 };
285
286 #define IP_REPLY_ARG_NOSRCCHECK 1
287
ip_reply_arg_flowi_flags(const struct ip_reply_arg * arg)288 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
289 {
290 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
291 }
292
293 void ip_send_unicast_reply(struct sock *sk, struct sk_buff *skb,
294 const struct ip_options *sopt,
295 __be32 daddr, __be32 saddr,
296 const struct ip_reply_arg *arg,
297 unsigned int len, u64 transmit_time, u32 txhash);
298
299 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
300 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
301 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
302 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
303 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
304 #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val)
305 #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
306 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
307 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
308 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
309
snmp_get_cpu_field(void __percpu * mib,int cpu,int offt)310 static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt)
311 {
312 return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt);
313 }
314
315 unsigned long snmp_fold_field(void __percpu *mib, int offt);
316 #if BITS_PER_LONG==32
317 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
318 size_t syncp_offset);
319 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
320 #else
snmp_get_cpu_field64(void __percpu * mib,int cpu,int offct,size_t syncp_offset)321 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
322 size_t syncp_offset)
323 {
324 return snmp_get_cpu_field(mib, cpu, offct);
325
326 }
327
snmp_fold_field64(void __percpu * mib,int offt,size_t syncp_off)328 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
329 {
330 return snmp_fold_field(mib, offt);
331 }
332 #endif
333
334 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
335 { \
336 int i, c; \
337 for_each_possible_cpu(c) { \
338 for (i = 0; stats_list[i].name; i++) \
339 buff64[i] += snmp_get_cpu_field64( \
340 mib_statistic, \
341 c, stats_list[i].entry, \
342 offset); \
343 } \
344 }
345
346 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
347 { \
348 int i, c; \
349 for_each_possible_cpu(c) { \
350 for (i = 0; stats_list[i].name; i++) \
351 buff[i] += snmp_get_cpu_field( \
352 mib_statistic, \
353 c, stats_list[i].entry); \
354 } \
355 }
356
inet_get_local_port_range(const struct net * net,int * low,int * high)357 static inline void inet_get_local_port_range(const struct net *net, int *low, int *high)
358 {
359 u32 range = READ_ONCE(net->ipv4.ip_local_ports.range);
360
361 *low = range & 0xffff;
362 *high = range >> 16;
363 }
364 bool inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high);
365
366 #ifdef CONFIG_SYSCTL
inet_is_local_reserved_port(struct net * net,unsigned short port)367 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
368 {
369 if (!net->ipv4.sysctl_local_reserved_ports)
370 return false;
371 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
372 }
373
sysctl_dev_name_is_allowed(const char * name)374 static inline bool sysctl_dev_name_is_allowed(const char *name)
375 {
376 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
377 }
378
inet_port_requires_bind_service(struct net * net,unsigned short port)379 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
380 {
381 return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
382 }
383
384 #else
inet_is_local_reserved_port(struct net * net,unsigned short port)385 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
386 {
387 return false;
388 }
389
inet_port_requires_bind_service(struct net * net,unsigned short port)390 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
391 {
392 return port < PROT_SOCK;
393 }
394 #endif
395
396 __be32 inet_current_timestamp(void);
397
398 /* From inetpeer.c */
399 extern int inet_peer_threshold;
400 extern int inet_peer_minttl;
401 extern int inet_peer_maxttl;
402
403 void ipfrag_init(void);
404
405 void ip_static_sysctl_init(void);
406
407 #define IP4_REPLY_MARK(net, mark) \
408 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
409
ip_is_fragment(const struct iphdr * iph)410 static inline bool ip_is_fragment(const struct iphdr *iph)
411 {
412 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
413 }
414
415 #ifdef CONFIG_INET
416 #include <net/dst.h>
417
418 /* The function in 2.2 was invalid, producing wrong result for
419 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
420 static inline
ip_decrease_ttl(struct iphdr * iph)421 int ip_decrease_ttl(struct iphdr *iph)
422 {
423 u32 check = (__force u32)iph->check;
424 check += (__force u32)htons(0x0100);
425 iph->check = (__force __sum16)(check + (check>=0xFFFF));
426 return --iph->ttl;
427 }
428
ip4h_dscp(const struct iphdr * ip4h)429 static inline dscp_t ip4h_dscp(const struct iphdr *ip4h)
430 {
431 return inet_dsfield_to_dscp(ip4h->tos);
432 }
433
ip_mtu_locked(const struct dst_entry * dst)434 static inline int ip_mtu_locked(const struct dst_entry *dst)
435 {
436 const struct rtable *rt = dst_rtable(dst);
437
438 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
439 }
440
441 static inline
ip_dont_fragment(const struct sock * sk,const struct dst_entry * dst)442 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
443 {
444 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
445
446 return pmtudisc == IP_PMTUDISC_DO ||
447 (pmtudisc == IP_PMTUDISC_WANT &&
448 !ip_mtu_locked(dst));
449 }
450
ip_sk_accept_pmtu(const struct sock * sk)451 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
452 {
453 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
454
455 return pmtudisc != IP_PMTUDISC_INTERFACE &&
456 pmtudisc != IP_PMTUDISC_OMIT;
457 }
458
ip_sk_use_pmtu(const struct sock * sk)459 static inline bool ip_sk_use_pmtu(const struct sock *sk)
460 {
461 return READ_ONCE(inet_sk(sk)->pmtudisc) < IP_PMTUDISC_PROBE;
462 }
463
ip_sk_ignore_df(const struct sock * sk)464 static inline bool ip_sk_ignore_df(const struct sock *sk)
465 {
466 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
467
468 return pmtudisc < IP_PMTUDISC_DO || pmtudisc == IP_PMTUDISC_OMIT;
469 }
470
ip_dst_mtu_maybe_forward(const struct dst_entry * dst,bool forwarding)471 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
472 bool forwarding)
473 {
474 const struct rtable *rt = dst_rtable(dst);
475 unsigned int mtu, res;
476 struct net *net;
477
478 rcu_read_lock();
479
480 net = dev_net_rcu(dst->dev);
481 if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
482 ip_mtu_locked(dst) ||
483 !forwarding) {
484 mtu = rt->rt_pmtu;
485 if (mtu && time_before(jiffies, rt->dst.expires))
486 goto out;
487 }
488
489 /* 'forwarding = true' case should always honour route mtu */
490 mtu = dst_metric_raw(dst, RTAX_MTU);
491 if (mtu)
492 goto out;
493
494 mtu = READ_ONCE(dst->dev->mtu);
495
496 if (unlikely(ip_mtu_locked(dst))) {
497 if (rt->rt_uses_gateway && mtu > 576)
498 mtu = 576;
499 }
500
501 out:
502 mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
503
504 res = mtu - lwtunnel_headroom(dst->lwtstate, mtu);
505
506 rcu_read_unlock();
507
508 return res;
509 }
510
ip_skb_dst_mtu(struct sock * sk,const struct sk_buff * skb)511 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
512 const struct sk_buff *skb)
513 {
514 unsigned int mtu;
515
516 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
517 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
518
519 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
520 }
521
522 mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
523 return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu);
524 }
525
526 struct dst_metrics *ip_fib_metrics_init(struct nlattr *fc_mx, int fc_mx_len,
527 struct netlink_ext_ack *extack);
ip_fib_metrics_put(struct dst_metrics * fib_metrics)528 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
529 {
530 if (fib_metrics != &dst_default_metrics &&
531 refcount_dec_and_test(&fib_metrics->refcnt))
532 kfree(fib_metrics);
533 }
534
535 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
536 static inline
ip_dst_init_metrics(struct dst_entry * dst,struct dst_metrics * fib_metrics)537 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
538 {
539 dst_init_metrics(dst, fib_metrics->metrics, true);
540
541 if (fib_metrics != &dst_default_metrics) {
542 dst->_metrics |= DST_METRICS_REFCOUNTED;
543 refcount_inc(&fib_metrics->refcnt);
544 }
545 }
546
547 static inline
ip_dst_metrics_put(struct dst_entry * dst)548 void ip_dst_metrics_put(struct dst_entry *dst)
549 {
550 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
551
552 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
553 kfree(p);
554 }
555
556 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
557
ip_select_ident_segs(struct net * net,struct sk_buff * skb,struct sock * sk,int segs)558 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
559 struct sock *sk, int segs)
560 {
561 struct iphdr *iph = ip_hdr(skb);
562
563 /* We had many attacks based on IPID, use the private
564 * generator as much as we can.
565 */
566 if (sk && inet_sk(sk)->inet_daddr) {
567 int val;
568
569 /* avoid atomic operations for TCP,
570 * as we hold socket lock at this point.
571 */
572 if (sk_is_tcp(sk)) {
573 sock_owned_by_me(sk);
574 val = atomic_read(&inet_sk(sk)->inet_id);
575 atomic_set(&inet_sk(sk)->inet_id, val + segs);
576 } else {
577 val = atomic_add_return(segs, &inet_sk(sk)->inet_id);
578 }
579 iph->id = htons(val);
580 return;
581 }
582 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
583 iph->id = 0;
584 } else {
585 /* Unfortunately we need the big hammer to get a suitable IPID */
586 __ip_select_ident(net, iph, segs);
587 }
588 }
589
ip_select_ident(struct net * net,struct sk_buff * skb,struct sock * sk)590 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
591 struct sock *sk)
592 {
593 ip_select_ident_segs(net, skb, sk, 1);
594 }
595
inet_compute_pseudo(struct sk_buff * skb,int proto)596 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
597 {
598 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
599 skb->len, proto, 0);
600 }
601
602 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
603 * Equivalent to : flow->v4addrs.src = iph->saddr;
604 * flow->v4addrs.dst = iph->daddr;
605 */
iph_to_flow_copy_v4addrs(struct flow_keys * flow,const struct iphdr * iph)606 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
607 const struct iphdr *iph)
608 {
609 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
610 offsetof(typeof(flow->addrs), v4addrs.src) +
611 sizeof(flow->addrs.v4addrs.src));
612 memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs));
613 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
614 }
615
616 /*
617 * Map a multicast IP onto multicast MAC for type ethernet.
618 */
619
ip_eth_mc_map(__be32 naddr,char * buf)620 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
621 {
622 __u32 addr=ntohl(naddr);
623 buf[0]=0x01;
624 buf[1]=0x00;
625 buf[2]=0x5e;
626 buf[5]=addr&0xFF;
627 addr>>=8;
628 buf[4]=addr&0xFF;
629 addr>>=8;
630 buf[3]=addr&0x7F;
631 }
632
633 /*
634 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
635 * Leave P_Key as 0 to be filled in by driver.
636 */
637
ip_ib_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)638 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
639 {
640 __u32 addr;
641 unsigned char scope = broadcast[5] & 0xF;
642
643 buf[0] = 0; /* Reserved */
644 buf[1] = 0xff; /* Multicast QPN */
645 buf[2] = 0xff;
646 buf[3] = 0xff;
647 addr = ntohl(naddr);
648 buf[4] = 0xff;
649 buf[5] = 0x10 | scope; /* scope from broadcast address */
650 buf[6] = 0x40; /* IPv4 signature */
651 buf[7] = 0x1b;
652 buf[8] = broadcast[8]; /* P_Key */
653 buf[9] = broadcast[9];
654 buf[10] = 0;
655 buf[11] = 0;
656 buf[12] = 0;
657 buf[13] = 0;
658 buf[14] = 0;
659 buf[15] = 0;
660 buf[19] = addr & 0xff;
661 addr >>= 8;
662 buf[18] = addr & 0xff;
663 addr >>= 8;
664 buf[17] = addr & 0xff;
665 addr >>= 8;
666 buf[16] = addr & 0x0f;
667 }
668
ip_ipgre_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)669 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
670 {
671 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
672 memcpy(buf, broadcast, 4);
673 else
674 memcpy(buf, &naddr, sizeof(naddr));
675 }
676
677 #if IS_ENABLED(CONFIG_IPV6)
678 #include <linux/ipv6.h>
679 #endif
680
inet_reset_saddr(struct sock * sk)681 static __inline__ void inet_reset_saddr(struct sock *sk)
682 {
683 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
684 #if IS_ENABLED(CONFIG_IPV6)
685 if (sk->sk_family == PF_INET6) {
686 struct ipv6_pinfo *np = inet6_sk(sk);
687
688 memset(&np->saddr, 0, sizeof(np->saddr));
689 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
690 }
691 #endif
692 }
693
694 #endif
695
ipv4_addr_hash(__be32 ip)696 static inline unsigned int ipv4_addr_hash(__be32 ip)
697 {
698 return (__force unsigned int) ip;
699 }
700
ipv4_portaddr_hash(const struct net * net,__be32 saddr,unsigned int port)701 static inline u32 ipv4_portaddr_hash(const struct net *net,
702 __be32 saddr,
703 unsigned int port)
704 {
705 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
706 }
707
708 bool ip_call_ra_chain(struct sk_buff *skb);
709
710 /*
711 * Functions provided by ip_fragment.c
712 */
713
714 enum ip_defrag_users {
715 IP_DEFRAG_LOCAL_DELIVER,
716 IP_DEFRAG_CALL_RA_CHAIN,
717 IP_DEFRAG_CONNTRACK_IN,
718 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
719 IP_DEFRAG_CONNTRACK_OUT,
720 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
721 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
722 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
723 IP_DEFRAG_VS_IN,
724 IP_DEFRAG_VS_OUT,
725 IP_DEFRAG_VS_FWD,
726 IP_DEFRAG_AF_PACKET,
727 IP_DEFRAG_MACVLAN,
728 };
729
730 /* Return true if the value of 'user' is between 'lower_bond'
731 * and 'upper_bond' inclusively.
732 */
ip_defrag_user_in_between(u32 user,enum ip_defrag_users lower_bond,enum ip_defrag_users upper_bond)733 static inline bool ip_defrag_user_in_between(u32 user,
734 enum ip_defrag_users lower_bond,
735 enum ip_defrag_users upper_bond)
736 {
737 return user >= lower_bond && user <= upper_bond;
738 }
739
740 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
741 #ifdef CONFIG_INET
742 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
743 #else
ip_check_defrag(struct net * net,struct sk_buff * skb,u32 user)744 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
745 {
746 return skb;
747 }
748 #endif
749
750 /*
751 * Functions provided by ip_forward.c
752 */
753
754 int ip_forward(struct sk_buff *skb);
755
756 /*
757 * Functions provided by ip_options.c
758 */
759
760 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
761 __be32 daddr, struct rtable *rt);
762
763 int __ip_options_echo(struct net *net, struct ip_options *dopt,
764 struct sk_buff *skb, const struct ip_options *sopt);
ip_options_echo(struct net * net,struct ip_options * dopt,struct sk_buff * skb)765 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
766 struct sk_buff *skb)
767 {
768 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
769 }
770
771 void ip_options_fragment(struct sk_buff *skb);
772 int __ip_options_compile(struct net *net, struct ip_options *opt,
773 struct sk_buff *skb, __be32 *info);
774 int ip_options_compile(struct net *net, struct ip_options *opt,
775 struct sk_buff *skb);
776 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
777 sockptr_t data, int optlen);
778 void ip_options_undo(struct ip_options *opt);
779 void ip_forward_options(struct sk_buff *skb);
780 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
781
782 /*
783 * Functions provided by ip_sockglue.c
784 */
785
786 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb, bool drop_dst);
787 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
788 struct sk_buff *skb, int tlen, int offset);
789 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
790 struct ipcm_cookie *ipc, bool allow_ipv6);
791 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl);
792 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
793 unsigned int optlen);
794 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
795 unsigned int optlen);
796 int do_ip_getsockopt(struct sock *sk, int level, int optname,
797 sockptr_t optval, sockptr_t optlen);
798 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
799 int __user *optlen);
800 int ip_ra_control(struct sock *sk, unsigned char on,
801 void (*destructor)(struct sock *));
802
803 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
804 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
805 u32 info, u8 *payload);
806 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
807 u32 info);
808
ip_cmsg_recv(struct msghdr * msg,struct sk_buff * skb)809 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
810 {
811 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
812 }
813
814 bool icmp_global_allow(struct net *net);
815 void icmp_global_consume(struct net *net);
816
817 #ifdef CONFIG_PROC_FS
818 int ip_misc_proc_init(void);
819 #endif
820
821 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
822 struct netlink_ext_ack *extack);
823
inetdev_valid_mtu(unsigned int mtu)824 static inline bool inetdev_valid_mtu(unsigned int mtu)
825 {
826 return likely(mtu >= IPV4_MIN_MTU);
827 }
828
829 void ip_sock_set_freebind(struct sock *sk);
830 int ip_sock_set_mtu_discover(struct sock *sk, int val);
831 void ip_sock_set_pktinfo(struct sock *sk);
832 void ip_sock_set_recverr(struct sock *sk);
833 void ip_sock_set_tos(struct sock *sk, int val);
834 void __ip_sock_set_tos(struct sock *sk, int val);
835
836 #endif /* _IP_H */
837