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
29 #include <net/inet_sock.h>
30 #include <net/route.h>
31 #include <net/snmp.h>
32 #include <net/flow.h>
33 #include <net/flow_dissector.h>
34 #include <net/netns/hash.h>
35 #include <net/lwtunnel.h>
36 #include <net/inet_dscp.h>
37
38 #define IPV4_MAX_PMTU 65535U /* RFC 2675, Section 5.1 */
39 #define IPV4_MIN_MTU 68 /* RFC 791 */
40
41 extern unsigned int sysctl_fib_sync_mem;
42 extern unsigned int sysctl_fib_sync_mem_min;
43 extern unsigned int sysctl_fib_sync_mem_max;
44
45 struct sock;
46
47 struct inet_skb_parm {
48 int iif;
49 struct ip_options opt; /* Compiled IP options */
50 u16 flags;
51
52 #define IPSKB_FORWARDED BIT(0)
53 #define IPSKB_XFRM_TUNNEL_SIZE BIT(1)
54 #define IPSKB_XFRM_TRANSFORMED BIT(2)
55 #define IPSKB_FRAG_COMPLETE BIT(3)
56 #define IPSKB_REROUTED BIT(4)
57 #define IPSKB_DOREDIRECT BIT(5)
58 #define IPSKB_FRAG_PMTU BIT(6)
59 #define IPSKB_L3SLAVE BIT(7)
60 #define IPSKB_NOPOLICY BIT(8)
61 #define IPSKB_MULTIPATH BIT(9)
62
63 u16 frag_max_size;
64 };
65
ipv4_l3mdev_skb(u16 flags)66 static inline bool ipv4_l3mdev_skb(u16 flags)
67 {
68 return !!(flags & IPSKB_L3SLAVE);
69 }
70
ip_hdrlen(const struct sk_buff * skb)71 static inline unsigned int ip_hdrlen(const struct sk_buff *skb)
72 {
73 return ip_hdr(skb)->ihl * 4;
74 }
75
76 struct ipcm_cookie {
77 struct sockcm_cookie sockc;
78 __be32 addr;
79 int oif;
80 struct ip_options_rcu *opt;
81 __u8 protocol;
82 __u8 ttl;
83 __s16 tos;
84 __u16 gso_size;
85 };
86
ipcm_init(struct ipcm_cookie * ipcm)87 static inline void ipcm_init(struct ipcm_cookie *ipcm)
88 {
89 *ipcm = (struct ipcm_cookie) { .tos = -1 };
90 }
91
ipcm_init_sk(struct ipcm_cookie * ipcm,const struct inet_sock * inet)92 static inline void ipcm_init_sk(struct ipcm_cookie *ipcm,
93 const struct inet_sock *inet)
94 {
95 ipcm_init(ipcm);
96
97 ipcm->sockc.mark = READ_ONCE(inet->sk.sk_mark);
98 ipcm->sockc.priority = READ_ONCE(inet->sk.sk_priority);
99 ipcm->sockc.tsflags = READ_ONCE(inet->sk.sk_tsflags);
100 ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if);
101 ipcm->addr = inet->inet_saddr;
102 ipcm->protocol = inet->inet_num;
103 }
104
105 #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb))
106 #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb))
107
108 /* return enslaved device index if relevant */
inet_sdif(const struct sk_buff * skb)109 static inline int inet_sdif(const struct sk_buff *skb)
110 {
111 #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV)
112 if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags))
113 return IPCB(skb)->iif;
114 #endif
115 return 0;
116 }
117
118 /* Special input handler for packets caught by router alert option.
119 They are selected only by protocol field, and then processed likely
120 local ones; but only if someone wants them! Otherwise, router
121 not running rsvpd will kill RSVP.
122
123 It is user level problem, what it will make with them.
124 I have no idea, how it will masquearde or NAT them (it is joke, joke :-)),
125 but receiver should be enough clever f.e. to forward mtrace requests,
126 sent to multicast group to reach destination designated router.
127 */
128
129 struct ip_ra_chain {
130 struct ip_ra_chain __rcu *next;
131 struct sock *sk;
132 union {
133 void (*destructor)(struct sock *);
134 struct sock *saved_sk;
135 };
136 struct rcu_head rcu;
137 };
138
139 /* IP flags. */
140 #define IP_CE 0x8000 /* Flag: "Congestion" */
141 #define IP_DF 0x4000 /* Flag: "Don't Fragment" */
142 #define IP_MF 0x2000 /* Flag: "More Fragments" */
143 #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */
144
145 #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */
146
147 struct msghdr;
148 struct net_device;
149 struct packet_type;
150 struct rtable;
151 struct sockaddr;
152
153 int igmp_mc_init(void);
154
155 /*
156 * Functions provided by ip.c
157 */
158
159 int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk,
160 __be32 saddr, __be32 daddr,
161 struct ip_options_rcu *opt, u8 tos);
162 int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt,
163 struct net_device *orig_dev);
164 void ip_list_rcv(struct list_head *head, struct packet_type *pt,
165 struct net_device *orig_dev);
166 int ip_local_deliver(struct sk_buff *skb);
167 void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto);
168 int ip_mr_input(struct sk_buff *skb);
169 int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb);
170 int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb);
171 int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
172 int (*output)(struct net *, struct sock *, struct sk_buff *));
173
174 struct ip_fraglist_iter {
175 struct sk_buff *frag;
176 struct iphdr *iph;
177 int offset;
178 unsigned int hlen;
179 };
180
181 void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph,
182 unsigned int hlen, struct ip_fraglist_iter *iter);
183 void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter);
184
ip_fraglist_next(struct ip_fraglist_iter * iter)185 static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter)
186 {
187 struct sk_buff *skb = iter->frag;
188
189 iter->frag = skb->next;
190 skb_mark_not_on_list(skb);
191
192 return skb;
193 }
194
195 struct ip_frag_state {
196 bool DF;
197 unsigned int hlen;
198 unsigned int ll_rs;
199 unsigned int mtu;
200 unsigned int left;
201 int offset;
202 int ptr;
203 __be16 not_last_frag;
204 };
205
206 void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs,
207 unsigned int mtu, bool DF, struct ip_frag_state *state);
208 struct sk_buff *ip_frag_next(struct sk_buff *skb,
209 struct ip_frag_state *state);
210
211 void ip_send_check(struct iphdr *ip);
212 int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
213 int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
214
215 int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
216 __u8 tos);
217 void ip_init(void);
218 int ip_append_data(struct sock *sk, struct flowi4 *fl4,
219 int getfrag(void *from, char *to, int offset, int len,
220 int odd, struct sk_buff *skb),
221 void *from, int len, int protolen,
222 struct ipcm_cookie *ipc,
223 struct rtable **rt,
224 unsigned int flags);
225 int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd,
226 struct sk_buff *skb);
227 struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4,
228 struct sk_buff_head *queue,
229 struct inet_cork *cork);
230 int ip_send_skb(struct net *net, struct sk_buff *skb);
231 int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4);
232 void ip_flush_pending_frames(struct sock *sk);
233 struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4,
234 int getfrag(void *from, char *to, int offset,
235 int len, int odd, struct sk_buff *skb),
236 void *from, int length, int transhdrlen,
237 struct ipcm_cookie *ipc, struct rtable **rtp,
238 struct inet_cork *cork, unsigned int flags);
239
240 int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl);
241
ip_finish_skb(struct sock * sk,struct flowi4 * fl4)242 static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4)
243 {
244 return __ip_make_skb(sk, fl4, &sk->sk_write_queue, &inet_sk(sk)->cork.base);
245 }
246
247 /* 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)248 static inline u8 ip_sendmsg_scope(const struct inet_sock *inet,
249 const struct ipcm_cookie *ipc,
250 const struct msghdr *msg)
251 {
252 if (sock_flag(&inet->sk, SOCK_LOCALROUTE) ||
253 msg->msg_flags & MSG_DONTROUTE ||
254 (ipc->opt && ipc->opt->opt.is_strictroute))
255 return RT_SCOPE_LINK;
256
257 return RT_SCOPE_UNIVERSE;
258 }
259
get_rttos(struct ipcm_cookie * ipc,struct inet_sock * inet)260 static inline __u8 get_rttos(struct ipcm_cookie* ipc, struct inet_sock *inet)
261 {
262 u8 dsfield = ipc->tos != -1 ? ipc->tos : READ_ONCE(inet->tos);
263
264 return dsfield & INET_DSCP_MASK;
265 }
266
267 /* datagram.c */
268 int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
269 int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len);
270
271 void ip4_datagram_release_cb(struct sock *sk);
272
273 struct ip_reply_arg {
274 struct kvec iov[1];
275 int flags;
276 __wsum csum;
277 int csumoffset; /* u16 offset of csum in iov[0].iov_base */
278 /* -1 if not needed */
279 int bound_dev_if;
280 u8 tos;
281 kuid_t uid;
282 };
283
284 #define IP_REPLY_ARG_NOSRCCHECK 1
285
ip_reply_arg_flowi_flags(const struct ip_reply_arg * arg)286 static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg)
287 {
288 return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0;
289 }
290
291 void ip_send_unicast_reply(struct sock *sk, const struct sock *orig_sk,
292 struct sk_buff *skb,
293 const struct ip_options *sopt,
294 __be32 daddr, __be32 saddr,
295 const struct ip_reply_arg *arg,
296 unsigned int len, u64 transmit_time, u32 txhash);
297
298 #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field)
299 #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field)
300 #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
301 #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val)
302 #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_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 NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field)
305 #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field)
306 #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
307 #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd)
308
snmp_get_cpu_field(void __percpu * mib,int cpu,int offt)309 static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt)
310 {
311 return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt);
312 }
313
314 unsigned long snmp_fold_field(void __percpu *mib, int offt);
315 #if BITS_PER_LONG==32
316 u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
317 size_t syncp_offset);
318 u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off);
319 #else
snmp_get_cpu_field64(void __percpu * mib,int cpu,int offct,size_t syncp_offset)320 static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct,
321 size_t syncp_offset)
322 {
323 return snmp_get_cpu_field(mib, cpu, offct);
324
325 }
326
snmp_fold_field64(void __percpu * mib,int offt,size_t syncp_off)327 static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off)
328 {
329 return snmp_fold_field(mib, offt);
330 }
331 #endif
332
333 #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \
334 { \
335 int i, c; \
336 for_each_possible_cpu(c) { \
337 for (i = 0; stats_list[i].name; i++) \
338 buff64[i] += snmp_get_cpu_field64( \
339 mib_statistic, \
340 c, stats_list[i].entry, \
341 offset); \
342 } \
343 }
344
345 #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \
346 { \
347 int i, c; \
348 for_each_possible_cpu(c) { \
349 for (i = 0; stats_list[i].name; i++) \
350 buff[i] += snmp_get_cpu_field( \
351 mib_statistic, \
352 c, stats_list[i].entry); \
353 } \
354 }
355
inet_get_local_port_range(const struct net * net,int * low,int * high)356 static inline void inet_get_local_port_range(const struct net *net, int *low, int *high)
357 {
358 u32 range = READ_ONCE(net->ipv4.ip_local_ports.range);
359
360 *low = range & 0xffff;
361 *high = range >> 16;
362 }
363 bool inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high);
364
365 #ifdef CONFIG_SYSCTL
inet_is_local_reserved_port(struct net * net,unsigned short port)366 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
367 {
368 if (!net->ipv4.sysctl_local_reserved_ports)
369 return false;
370 return test_bit(port, net->ipv4.sysctl_local_reserved_ports);
371 }
372
sysctl_dev_name_is_allowed(const char * name)373 static inline bool sysctl_dev_name_is_allowed(const char *name)
374 {
375 return strcmp(name, "default") != 0 && strcmp(name, "all") != 0;
376 }
377
inet_port_requires_bind_service(struct net * net,unsigned short port)378 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
379 {
380 return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock);
381 }
382
383 #else
inet_is_local_reserved_port(struct net * net,unsigned short port)384 static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port)
385 {
386 return false;
387 }
388
inet_port_requires_bind_service(struct net * net,unsigned short port)389 static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port)
390 {
391 return port < PROT_SOCK;
392 }
393 #endif
394
395 __be32 inet_current_timestamp(void);
396
397 /* From inetpeer.c */
398 extern int inet_peer_threshold;
399 extern int inet_peer_minttl;
400 extern int inet_peer_maxttl;
401
402 void ipfrag_init(void);
403
404 void ip_static_sysctl_init(void);
405
406 #define IP4_REPLY_MARK(net, mark) \
407 (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0)
408
ip_is_fragment(const struct iphdr * iph)409 static inline bool ip_is_fragment(const struct iphdr *iph)
410 {
411 return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0;
412 }
413
414 #ifdef CONFIG_INET
415 #include <net/dst.h>
416
417 /* The function in 2.2 was invalid, producing wrong result for
418 * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */
419 static inline
ip_decrease_ttl(struct iphdr * iph)420 int ip_decrease_ttl(struct iphdr *iph)
421 {
422 u32 check = (__force u32)iph->check;
423 check += (__force u32)htons(0x0100);
424 iph->check = (__force __sum16)(check + (check>=0xFFFF));
425 return --iph->ttl;
426 }
427
ip4h_dscp(const struct iphdr * ip4h)428 static inline dscp_t ip4h_dscp(const struct iphdr *ip4h)
429 {
430 return inet_dsfield_to_dscp(ip4h->tos);
431 }
432
ip_mtu_locked(const struct dst_entry * dst)433 static inline int ip_mtu_locked(const struct dst_entry *dst)
434 {
435 const struct rtable *rt = dst_rtable(dst);
436
437 return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU);
438 }
439
440 static inline
ip_dont_fragment(const struct sock * sk,const struct dst_entry * dst)441 int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst)
442 {
443 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
444
445 return pmtudisc == IP_PMTUDISC_DO ||
446 (pmtudisc == IP_PMTUDISC_WANT &&
447 !ip_mtu_locked(dst));
448 }
449
ip_sk_accept_pmtu(const struct sock * sk)450 static inline bool ip_sk_accept_pmtu(const struct sock *sk)
451 {
452 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
453
454 return pmtudisc != IP_PMTUDISC_INTERFACE &&
455 pmtudisc != IP_PMTUDISC_OMIT;
456 }
457
ip_sk_use_pmtu(const struct sock * sk)458 static inline bool ip_sk_use_pmtu(const struct sock *sk)
459 {
460 return READ_ONCE(inet_sk(sk)->pmtudisc) < IP_PMTUDISC_PROBE;
461 }
462
ip_sk_ignore_df(const struct sock * sk)463 static inline bool ip_sk_ignore_df(const struct sock *sk)
464 {
465 u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc);
466
467 return pmtudisc < IP_PMTUDISC_DO || pmtudisc == IP_PMTUDISC_OMIT;
468 }
469
ip_dst_mtu_maybe_forward(const struct dst_entry * dst,bool forwarding)470 static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst,
471 bool forwarding)
472 {
473 const struct rtable *rt = dst_rtable(dst);
474 unsigned int mtu, res;
475 struct net *net;
476
477 rcu_read_lock();
478
479 net = dev_net_rcu(dst->dev);
480 if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) ||
481 ip_mtu_locked(dst) ||
482 !forwarding) {
483 mtu = rt->rt_pmtu;
484 if (mtu && time_before(jiffies, rt->dst.expires))
485 goto out;
486 }
487
488 /* 'forwarding = true' case should always honour route mtu */
489 mtu = dst_metric_raw(dst, RTAX_MTU);
490 if (mtu)
491 goto out;
492
493 mtu = READ_ONCE(dst->dev->mtu);
494
495 if (unlikely(ip_mtu_locked(dst))) {
496 if (rt->rt_uses_gateway && mtu > 576)
497 mtu = 576;
498 }
499
500 out:
501 mtu = min_t(unsigned int, mtu, IP_MAX_MTU);
502
503 res = mtu - lwtunnel_headroom(dst->lwtstate, mtu);
504
505 rcu_read_unlock();
506
507 return res;
508 }
509
ip_skb_dst_mtu(struct sock * sk,const struct sk_buff * skb)510 static inline unsigned int ip_skb_dst_mtu(struct sock *sk,
511 const struct sk_buff *skb)
512 {
513 unsigned int mtu;
514
515 if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) {
516 bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED;
517
518 return ip_dst_mtu_maybe_forward(skb_dst(skb), forwarding);
519 }
520
521 mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU);
522 return mtu - lwtunnel_headroom(skb_dst(skb)->lwtstate, mtu);
523 }
524
525 struct dst_metrics *ip_fib_metrics_init(struct nlattr *fc_mx, int fc_mx_len,
526 struct netlink_ext_ack *extack);
ip_fib_metrics_put(struct dst_metrics * fib_metrics)527 static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics)
528 {
529 if (fib_metrics != &dst_default_metrics &&
530 refcount_dec_and_test(&fib_metrics->refcnt))
531 kfree(fib_metrics);
532 }
533
534 /* ipv4 and ipv6 both use refcounted metrics if it is not the default */
535 static inline
ip_dst_init_metrics(struct dst_entry * dst,struct dst_metrics * fib_metrics)536 void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics)
537 {
538 dst_init_metrics(dst, fib_metrics->metrics, true);
539
540 if (fib_metrics != &dst_default_metrics) {
541 dst->_metrics |= DST_METRICS_REFCOUNTED;
542 refcount_inc(&fib_metrics->refcnt);
543 }
544 }
545
546 static inline
ip_dst_metrics_put(struct dst_entry * dst)547 void ip_dst_metrics_put(struct dst_entry *dst)
548 {
549 struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst);
550
551 if (p != &dst_default_metrics && refcount_dec_and_test(&p->refcnt))
552 kfree(p);
553 }
554
555 void __ip_select_ident(struct net *net, struct iphdr *iph, int segs);
556
ip_select_ident_segs(struct net * net,struct sk_buff * skb,struct sock * sk,int segs)557 static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb,
558 struct sock *sk, int segs)
559 {
560 struct iphdr *iph = ip_hdr(skb);
561
562 /* We had many attacks based on IPID, use the private
563 * generator as much as we can.
564 */
565 if (sk && inet_sk(sk)->inet_daddr) {
566 int val;
567
568 /* avoid atomic operations for TCP,
569 * as we hold socket lock at this point.
570 */
571 if (sk_is_tcp(sk)) {
572 sock_owned_by_me(sk);
573 val = atomic_read(&inet_sk(sk)->inet_id);
574 atomic_set(&inet_sk(sk)->inet_id, val + segs);
575 } else {
576 val = atomic_add_return(segs, &inet_sk(sk)->inet_id);
577 }
578 iph->id = htons(val);
579 return;
580 }
581 if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) {
582 iph->id = 0;
583 } else {
584 /* Unfortunately we need the big hammer to get a suitable IPID */
585 __ip_select_ident(net, iph, segs);
586 }
587 }
588
ip_select_ident(struct net * net,struct sk_buff * skb,struct sock * sk)589 static inline void ip_select_ident(struct net *net, struct sk_buff *skb,
590 struct sock *sk)
591 {
592 ip_select_ident_segs(net, skb, sk, 1);
593 }
594
inet_compute_pseudo(struct sk_buff * skb,int proto)595 static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto)
596 {
597 return csum_tcpudp_nofold(ip_hdr(skb)->saddr, ip_hdr(skb)->daddr,
598 skb->len, proto, 0);
599 }
600
601 /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store
602 * Equivalent to : flow->v4addrs.src = iph->saddr;
603 * flow->v4addrs.dst = iph->daddr;
604 */
iph_to_flow_copy_v4addrs(struct flow_keys * flow,const struct iphdr * iph)605 static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow,
606 const struct iphdr *iph)
607 {
608 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) !=
609 offsetof(typeof(flow->addrs), v4addrs.src) +
610 sizeof(flow->addrs.v4addrs.src));
611 memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs));
612 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
613 }
614
615 /*
616 * Map a multicast IP onto multicast MAC for type ethernet.
617 */
618
ip_eth_mc_map(__be32 naddr,char * buf)619 static inline void ip_eth_mc_map(__be32 naddr, char *buf)
620 {
621 __u32 addr=ntohl(naddr);
622 buf[0]=0x01;
623 buf[1]=0x00;
624 buf[2]=0x5e;
625 buf[5]=addr&0xFF;
626 addr>>=8;
627 buf[4]=addr&0xFF;
628 addr>>=8;
629 buf[3]=addr&0x7F;
630 }
631
632 /*
633 * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand.
634 * Leave P_Key as 0 to be filled in by driver.
635 */
636
ip_ib_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)637 static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
638 {
639 __u32 addr;
640 unsigned char scope = broadcast[5] & 0xF;
641
642 buf[0] = 0; /* Reserved */
643 buf[1] = 0xff; /* Multicast QPN */
644 buf[2] = 0xff;
645 buf[3] = 0xff;
646 addr = ntohl(naddr);
647 buf[4] = 0xff;
648 buf[5] = 0x10 | scope; /* scope from broadcast address */
649 buf[6] = 0x40; /* IPv4 signature */
650 buf[7] = 0x1b;
651 buf[8] = broadcast[8]; /* P_Key */
652 buf[9] = broadcast[9];
653 buf[10] = 0;
654 buf[11] = 0;
655 buf[12] = 0;
656 buf[13] = 0;
657 buf[14] = 0;
658 buf[15] = 0;
659 buf[19] = addr & 0xff;
660 addr >>= 8;
661 buf[18] = addr & 0xff;
662 addr >>= 8;
663 buf[17] = addr & 0xff;
664 addr >>= 8;
665 buf[16] = addr & 0x0f;
666 }
667
ip_ipgre_mc_map(__be32 naddr,const unsigned char * broadcast,char * buf)668 static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf)
669 {
670 if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0)
671 memcpy(buf, broadcast, 4);
672 else
673 memcpy(buf, &naddr, sizeof(naddr));
674 }
675
676 #if IS_ENABLED(CONFIG_IPV6)
677 #include <linux/ipv6.h>
678 #endif
679
inet_reset_saddr(struct sock * sk)680 static __inline__ void inet_reset_saddr(struct sock *sk)
681 {
682 inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0;
683 #if IS_ENABLED(CONFIG_IPV6)
684 if (sk->sk_family == PF_INET6) {
685 struct ipv6_pinfo *np = inet6_sk(sk);
686
687 memset(&np->saddr, 0, sizeof(np->saddr));
688 memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr));
689 }
690 #endif
691 }
692
693 #endif
694
ipv4_addr_hash(__be32 ip)695 static inline unsigned int ipv4_addr_hash(__be32 ip)
696 {
697 return (__force unsigned int) ip;
698 }
699
__ipv4_addr_hash(const __be32 ip,const u32 initval)700 static inline u32 __ipv4_addr_hash(const __be32 ip, const u32 initval)
701 {
702 return jhash_1word((__force u32)ip, initval);
703 }
704
ipv4_portaddr_hash(const struct net * net,__be32 saddr,unsigned int port)705 static inline u32 ipv4_portaddr_hash(const struct net *net,
706 __be32 saddr,
707 unsigned int port)
708 {
709 return jhash_1word((__force u32)saddr, net_hash_mix(net)) ^ port;
710 }
711
712 bool ip_call_ra_chain(struct sk_buff *skb);
713
714 /*
715 * Functions provided by ip_fragment.c
716 */
717
718 enum ip_defrag_users {
719 IP_DEFRAG_LOCAL_DELIVER,
720 IP_DEFRAG_CALL_RA_CHAIN,
721 IP_DEFRAG_CONNTRACK_IN,
722 __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX,
723 IP_DEFRAG_CONNTRACK_OUT,
724 __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX,
725 IP_DEFRAG_CONNTRACK_BRIDGE_IN,
726 __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX,
727 IP_DEFRAG_VS_IN,
728 IP_DEFRAG_VS_OUT,
729 IP_DEFRAG_VS_FWD,
730 IP_DEFRAG_AF_PACKET,
731 IP_DEFRAG_MACVLAN,
732 };
733
734 /* Return true if the value of 'user' is between 'lower_bond'
735 * and 'upper_bond' inclusively.
736 */
ip_defrag_user_in_between(u32 user,enum ip_defrag_users lower_bond,enum ip_defrag_users upper_bond)737 static inline bool ip_defrag_user_in_between(u32 user,
738 enum ip_defrag_users lower_bond,
739 enum ip_defrag_users upper_bond)
740 {
741 return user >= lower_bond && user <= upper_bond;
742 }
743
744 int ip_defrag(struct net *net, struct sk_buff *skb, u32 user);
745 #ifdef CONFIG_INET
746 struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user);
747 #else
ip_check_defrag(struct net * net,struct sk_buff * skb,u32 user)748 static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user)
749 {
750 return skb;
751 }
752 #endif
753
754 /*
755 * Functions provided by ip_forward.c
756 */
757
758 int ip_forward(struct sk_buff *skb);
759
760 /*
761 * Functions provided by ip_options.c
762 */
763
764 void ip_options_build(struct sk_buff *skb, struct ip_options *opt,
765 __be32 daddr, struct rtable *rt);
766
767 int __ip_options_echo(struct net *net, struct ip_options *dopt,
768 struct sk_buff *skb, const struct ip_options *sopt);
ip_options_echo(struct net * net,struct ip_options * dopt,struct sk_buff * skb)769 static inline int ip_options_echo(struct net *net, struct ip_options *dopt,
770 struct sk_buff *skb)
771 {
772 return __ip_options_echo(net, dopt, skb, &IPCB(skb)->opt);
773 }
774
775 void ip_options_fragment(struct sk_buff *skb);
776 int __ip_options_compile(struct net *net, struct ip_options *opt,
777 struct sk_buff *skb, __be32 *info);
778 int ip_options_compile(struct net *net, struct ip_options *opt,
779 struct sk_buff *skb);
780 int ip_options_get(struct net *net, struct ip_options_rcu **optp,
781 sockptr_t data, int optlen);
782 void ip_options_undo(struct ip_options *opt);
783 void ip_forward_options(struct sk_buff *skb);
784 int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev);
785
786 /*
787 * Functions provided by ip_sockglue.c
788 */
789
790 void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb, bool drop_dst);
791 void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk,
792 struct sk_buff *skb, int tlen, int offset);
793 int ip_cmsg_send(struct sock *sk, struct msghdr *msg,
794 struct ipcm_cookie *ipc, bool allow_ipv6);
795 DECLARE_STATIC_KEY_FALSE(ip4_min_ttl);
796 int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
797 unsigned int optlen);
798 int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
799 unsigned int optlen);
800 int do_ip_getsockopt(struct sock *sk, int level, int optname,
801 sockptr_t optval, sockptr_t optlen);
802 int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
803 int __user *optlen);
804 int ip_ra_control(struct sock *sk, unsigned char on,
805 void (*destructor)(struct sock *));
806
807 int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len);
808 void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
809 u32 info, u8 *payload);
810 void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport,
811 u32 info);
812
ip_cmsg_recv(struct msghdr * msg,struct sk_buff * skb)813 static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb)
814 {
815 ip_cmsg_recv_offset(msg, skb->sk, skb, 0, 0);
816 }
817
818 bool icmp_global_allow(struct net *net);
819 void icmp_global_consume(struct net *net);
820
821 #ifdef CONFIG_PROC_FS
822 int ip_misc_proc_init(void);
823 #endif
824
825 int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family,
826 struct netlink_ext_ack *extack);
827
inetdev_valid_mtu(unsigned int mtu)828 static inline bool inetdev_valid_mtu(unsigned int mtu)
829 {
830 return likely(mtu >= IPV4_MIN_MTU);
831 }
832
833 void ip_sock_set_freebind(struct sock *sk);
834 int ip_sock_set_mtu_discover(struct sock *sk, int val);
835 void ip_sock_set_pktinfo(struct sock *sk);
836 void ip_sock_set_recverr(struct sock *sk);
837 void ip_sock_set_tos(struct sock *sk, int val);
838 void __ip_sock_set_tos(struct sock *sk, int val);
839
840 #endif /* _IP_H */
841