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 | |
66 | static inline bool ipv4_l3mdev_skb(u16 flags) |
67 | { |
68 | return !!(flags & IPSKB_L3SLAVE); |
69 | } |
70 | |
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 | |
87 | static inline void ipcm_init(struct ipcm_cookie *ipcm) |
88 | { |
89 | *ipcm = (struct ipcm_cookie) { .tos = -1 }; |
90 | } |
91 | |
92 | static inline void ipcm_init_sk(struct ipcm_cookie *ipcm, |
93 | const struct inet_sock *inet) |
94 | { |
95 | *ipcm = (struct ipcm_cookie) { |
96 | .tos = READ_ONCE(inet->tos), |
97 | }; |
98 | |
99 | sockcm_init(sockc: &ipcm->sockc, sk: &inet->sk); |
100 | |
101 | ipcm->oif = READ_ONCE(inet->sk.sk_bound_dev_if); |
102 | ipcm->addr = inet->inet_saddr; |
103 | ipcm->protocol = inet->inet_num; |
104 | } |
105 | |
106 | #define IPCB(skb) ((struct inet_skb_parm*)((skb)->cb)) |
107 | #define PKTINFO_SKB_CB(skb) ((struct in_pktinfo *)((skb)->cb)) |
108 | |
109 | /* return enslaved device index if relevant */ |
110 | static inline int inet_sdif(const struct sk_buff *skb) |
111 | { |
112 | #if IS_ENABLED(CONFIG_NET_L3_MASTER_DEV) |
113 | if (skb && ipv4_l3mdev_skb(IPCB(skb)->flags)) |
114 | return IPCB(skb)->iif; |
115 | #endif |
116 | return 0; |
117 | } |
118 | |
119 | /* Special input handler for packets caught by router alert option. |
120 | They are selected only by protocol field, and then processed likely |
121 | local ones; but only if someone wants them! Otherwise, router |
122 | not running rsvpd will kill RSVP. |
123 | |
124 | It is user level problem, what it will make with them. |
125 | I have no idea, how it will masquearde or NAT them (it is joke, joke :-)), |
126 | but receiver should be enough clever f.e. to forward mtrace requests, |
127 | sent to multicast group to reach destination designated router. |
128 | */ |
129 | |
130 | struct ip_ra_chain { |
131 | struct ip_ra_chain __rcu *next; |
132 | struct sock *sk; |
133 | union { |
134 | void (*destructor)(struct sock *); |
135 | struct sock *saved_sk; |
136 | }; |
137 | struct rcu_head rcu; |
138 | }; |
139 | |
140 | /* IP flags. */ |
141 | #define IP_CE 0x8000 /* Flag: "Congestion" */ |
142 | #define IP_DF 0x4000 /* Flag: "Don't Fragment" */ |
143 | #define IP_MF 0x2000 /* Flag: "More Fragments" */ |
144 | #define IP_OFFSET 0x1FFF /* "Fragment Offset" part */ |
145 | |
146 | #define IP_FRAG_TIME (30 * HZ) /* fragment lifetime */ |
147 | |
148 | struct msghdr; |
149 | struct net_device; |
150 | struct packet_type; |
151 | struct rtable; |
152 | struct sockaddr; |
153 | |
154 | int igmp_mc_init(void); |
155 | |
156 | /* |
157 | * Functions provided by ip.c |
158 | */ |
159 | |
160 | int ip_build_and_send_pkt(struct sk_buff *skb, const struct sock *sk, |
161 | __be32 saddr, __be32 daddr, |
162 | struct ip_options_rcu *opt, u8 tos); |
163 | int ip_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, |
164 | struct net_device *orig_dev); |
165 | void ip_list_rcv(struct list_head *head, struct packet_type *pt, |
166 | struct net_device *orig_dev); |
167 | int ip_local_deliver(struct sk_buff *skb); |
168 | void ip_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int proto); |
169 | int ip_mr_input(struct sk_buff *skb); |
170 | int ip_output(struct net *net, struct sock *sk, struct sk_buff *skb); |
171 | int ip_mc_output(struct net *net, struct sock *sk, struct sk_buff *skb); |
172 | int ip_do_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, |
173 | int (*output)(struct net *, struct sock *, struct sk_buff *)); |
174 | |
175 | struct ip_fraglist_iter { |
176 | struct sk_buff *frag; |
177 | struct iphdr *iph; |
178 | int offset; |
179 | unsigned int hlen; |
180 | }; |
181 | |
182 | void ip_fraglist_init(struct sk_buff *skb, struct iphdr *iph, |
183 | unsigned int hlen, struct ip_fraglist_iter *iter); |
184 | void ip_fraglist_prepare(struct sk_buff *skb, struct ip_fraglist_iter *iter); |
185 | |
186 | static inline struct sk_buff *ip_fraglist_next(struct ip_fraglist_iter *iter) |
187 | { |
188 | struct sk_buff *skb = iter->frag; |
189 | |
190 | iter->frag = skb->next; |
191 | skb_mark_not_on_list(skb); |
192 | |
193 | return skb; |
194 | } |
195 | |
196 | struct ip_frag_state { |
197 | bool DF; |
198 | unsigned int hlen; |
199 | unsigned int ll_rs; |
200 | unsigned int mtu; |
201 | unsigned int left; |
202 | int offset; |
203 | int ptr; |
204 | __be16 not_last_frag; |
205 | }; |
206 | |
207 | void ip_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int ll_rs, |
208 | unsigned int mtu, bool DF, struct ip_frag_state *state); |
209 | struct sk_buff *ip_frag_next(struct sk_buff *skb, |
210 | struct ip_frag_state *state); |
211 | |
212 | void ip_send_check(struct iphdr *ip); |
213 | int __ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); |
214 | int ip_local_out(struct net *net, struct sock *sk, struct sk_buff *skb); |
215 | |
216 | int __ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl, |
217 | __u8 tos); |
218 | void ip_init(void); |
219 | int ip_append_data(struct sock *sk, struct flowi4 *fl4, |
220 | int getfrag(void *from, char *to, int offset, int len, |
221 | int odd, struct sk_buff *skb), |
222 | void *from, int len, int protolen, |
223 | struct ipcm_cookie *ipc, |
224 | struct rtable **rt, |
225 | unsigned int flags); |
226 | int ip_generic_getfrag(void *from, char *to, int offset, int len, int odd, |
227 | struct sk_buff *skb); |
228 | struct sk_buff *__ip_make_skb(struct sock *sk, struct flowi4 *fl4, |
229 | struct sk_buff_head *queue, |
230 | struct inet_cork *cork); |
231 | int ip_send_skb(struct net *net, struct sk_buff *skb); |
232 | int ip_push_pending_frames(struct sock *sk, struct flowi4 *fl4); |
233 | void ip_flush_pending_frames(struct sock *sk); |
234 | struct sk_buff *ip_make_skb(struct sock *sk, struct flowi4 *fl4, |
235 | int getfrag(void *from, char *to, int offset, |
236 | int len, int odd, struct sk_buff *skb), |
237 | void *from, int length, int transhdrlen, |
238 | struct ipcm_cookie *ipc, struct rtable **rtp, |
239 | struct inet_cork *cork, unsigned int flags); |
240 | |
241 | int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl); |
242 | |
243 | static inline struct sk_buff *ip_finish_skb(struct sock *sk, struct flowi4 *fl4) |
244 | { |
245 | return __ip_make_skb(sk, fl4, queue: &sk->sk_write_queue, cork: &inet_sk(sk)->cork.base); |
246 | } |
247 | |
248 | /* Get the route scope that should be used when sending a packet. */ |
249 | static inline u8 ip_sendmsg_scope(const struct inet_sock *inet, |
250 | const struct ipcm_cookie *ipc, |
251 | const struct msghdr *msg) |
252 | { |
253 | if (sock_flag(sk: &inet->sk, flag: SOCK_LOCALROUTE) || |
254 | msg->msg_flags & MSG_DONTROUTE || |
255 | (ipc->opt && ipc->opt->opt.is_strictroute)) |
256 | return RT_SCOPE_LINK; |
257 | |
258 | return RT_SCOPE_UNIVERSE; |
259 | } |
260 | |
261 | /* datagram.c */ |
262 | int __ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); |
263 | int ip4_datagram_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len); |
264 | |
265 | void ip4_datagram_release_cb(struct sock *sk); |
266 | |
267 | struct ip_reply_arg { |
268 | struct kvec iov[1]; |
269 | int flags; |
270 | __wsum csum; |
271 | int csumoffset; /* u16 offset of csum in iov[0].iov_base */ |
272 | /* -1 if not needed */ |
273 | int bound_dev_if; |
274 | u8 tos; |
275 | kuid_t uid; |
276 | }; |
277 | |
278 | #define IP_REPLY_ARG_NOSRCCHECK 1 |
279 | |
280 | static inline __u8 ip_reply_arg_flowi_flags(const struct ip_reply_arg *arg) |
281 | { |
282 | return (arg->flags & IP_REPLY_ARG_NOSRCCHECK) ? FLOWI_FLAG_ANYSRC : 0; |
283 | } |
284 | |
285 | void ip_send_unicast_reply(struct sock *sk, const struct sock *orig_sk, |
286 | struct sk_buff *skb, |
287 | const struct ip_options *sopt, |
288 | __be32 daddr, __be32 saddr, |
289 | const struct ip_reply_arg *arg, |
290 | unsigned int len, u64 transmit_time, u32 txhash); |
291 | |
292 | #define IP_INC_STATS(net, field) SNMP_INC_STATS64((net)->mib.ip_statistics, field) |
293 | #define __IP_INC_STATS(net, field) __SNMP_INC_STATS64((net)->mib.ip_statistics, field) |
294 | #define IP_ADD_STATS(net, field, val) SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) |
295 | #define __IP_ADD_STATS(net, field, val) __SNMP_ADD_STATS64((net)->mib.ip_statistics, field, val) |
296 | #define IP_UPD_PO_STATS(net, field, val) SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) |
297 | #define __IP_UPD_PO_STATS(net, field, val) __SNMP_UPD_PO_STATS64((net)->mib.ip_statistics, field, val) |
298 | #define NET_INC_STATS(net, field) SNMP_INC_STATS((net)->mib.net_statistics, field) |
299 | #define __NET_INC_STATS(net, field) __SNMP_INC_STATS((net)->mib.net_statistics, field) |
300 | #define NET_ADD_STATS(net, field, adnd) SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) |
301 | #define __NET_ADD_STATS(net, field, adnd) __SNMP_ADD_STATS((net)->mib.net_statistics, field, adnd) |
302 | |
303 | static inline u64 snmp_get_cpu_field(void __percpu *mib, int cpu, int offt) |
304 | { |
305 | return *(((unsigned long *)per_cpu_ptr(mib, cpu)) + offt); |
306 | } |
307 | |
308 | unsigned long snmp_fold_field(void __percpu *mib, int offt); |
309 | #if BITS_PER_LONG==32 |
310 | u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, |
311 | size_t syncp_offset); |
312 | u64 snmp_fold_field64(void __percpu *mib, int offt, size_t sync_off); |
313 | #else |
314 | static inline u64 snmp_get_cpu_field64(void __percpu *mib, int cpu, int offct, |
315 | size_t syncp_offset) |
316 | { |
317 | return snmp_get_cpu_field(mib, cpu, offt: offct); |
318 | |
319 | } |
320 | |
321 | static inline u64 snmp_fold_field64(void __percpu *mib, int offt, size_t syncp_off) |
322 | { |
323 | return snmp_fold_field(mib, offt); |
324 | } |
325 | #endif |
326 | |
327 | #define snmp_get_cpu_field64_batch(buff64, stats_list, mib_statistic, offset) \ |
328 | { \ |
329 | int i, c; \ |
330 | for_each_possible_cpu(c) { \ |
331 | for (i = 0; stats_list[i].name; i++) \ |
332 | buff64[i] += snmp_get_cpu_field64( \ |
333 | mib_statistic, \ |
334 | c, stats_list[i].entry, \ |
335 | offset); \ |
336 | } \ |
337 | } |
338 | |
339 | #define snmp_get_cpu_field_batch(buff, stats_list, mib_statistic) \ |
340 | { \ |
341 | int i, c; \ |
342 | for_each_possible_cpu(c) { \ |
343 | for (i = 0; stats_list[i].name; i++) \ |
344 | buff[i] += snmp_get_cpu_field( \ |
345 | mib_statistic, \ |
346 | c, stats_list[i].entry); \ |
347 | } \ |
348 | } |
349 | |
350 | static inline void inet_get_local_port_range(const struct net *net, int *low, int *high) |
351 | { |
352 | u32 range = READ_ONCE(net->ipv4.ip_local_ports.range); |
353 | |
354 | *low = range & 0xffff; |
355 | *high = range >> 16; |
356 | } |
357 | bool inet_sk_get_local_port_range(const struct sock *sk, int *low, int *high); |
358 | |
359 | #ifdef CONFIG_SYSCTL |
360 | static inline bool inet_is_local_reserved_port(const struct net *net, unsigned short port) |
361 | { |
362 | if (!net->ipv4.sysctl_local_reserved_ports) |
363 | return false; |
364 | return test_bit(port, net->ipv4.sysctl_local_reserved_ports); |
365 | } |
366 | |
367 | static inline bool sysctl_dev_name_is_allowed(const char *name) |
368 | { |
369 | return strcmp(name, "default") != 0 && strcmp(name, "all") != 0; |
370 | } |
371 | |
372 | static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) |
373 | { |
374 | return port < READ_ONCE(net->ipv4.sysctl_ip_prot_sock); |
375 | } |
376 | |
377 | #else |
378 | static inline bool inet_is_local_reserved_port(struct net *net, unsigned short port) |
379 | { |
380 | return false; |
381 | } |
382 | |
383 | static inline bool inet_port_requires_bind_service(struct net *net, unsigned short port) |
384 | { |
385 | return port < PROT_SOCK; |
386 | } |
387 | #endif |
388 | |
389 | __be32 inet_current_timestamp(void); |
390 | |
391 | /* From inetpeer.c */ |
392 | extern int inet_peer_threshold; |
393 | extern int inet_peer_minttl; |
394 | extern int inet_peer_maxttl; |
395 | |
396 | void ipfrag_init(void); |
397 | |
398 | void ip_static_sysctl_init(void); |
399 | |
400 | #define IP4_REPLY_MARK(net, mark) \ |
401 | (READ_ONCE((net)->ipv4.sysctl_fwmark_reflect) ? (mark) : 0) |
402 | |
403 | static inline bool ip_is_fragment(const struct iphdr *iph) |
404 | { |
405 | return (iph->frag_off & htons(IP_MF | IP_OFFSET)) != 0; |
406 | } |
407 | |
408 | #ifdef CONFIG_INET |
409 | #include <net/dst.h> |
410 | |
411 | /* The function in 2.2 was invalid, producing wrong result for |
412 | * check=0xFEFF. It was noticed by Arthur Skawina _year_ ago. --ANK(000625) */ |
413 | static inline |
414 | int ip_decrease_ttl(struct iphdr *iph) |
415 | { |
416 | u32 check = (__force u32)iph->check; |
417 | check += (__force u32)htons(0x0100); |
418 | iph->check = (__force __sum16)(check + (check>=0xFFFF)); |
419 | return --iph->ttl; |
420 | } |
421 | |
422 | static inline dscp_t ip4h_dscp(const struct iphdr *ip4h) |
423 | { |
424 | return inet_dsfield_to_dscp(dsfield: ip4h->tos); |
425 | } |
426 | |
427 | static inline int ip_mtu_locked(const struct dst_entry *dst) |
428 | { |
429 | const struct rtable *rt = dst_rtable(dst); |
430 | |
431 | return rt->rt_mtu_locked || dst_metric_locked(dst, RTAX_MTU); |
432 | } |
433 | |
434 | static inline |
435 | int ip_dont_fragment(const struct sock *sk, const struct dst_entry *dst) |
436 | { |
437 | u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); |
438 | |
439 | return pmtudisc == IP_PMTUDISC_DO || |
440 | (pmtudisc == IP_PMTUDISC_WANT && |
441 | !ip_mtu_locked(dst)); |
442 | } |
443 | |
444 | static inline bool ip_sk_accept_pmtu(const struct sock *sk) |
445 | { |
446 | u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); |
447 | |
448 | return pmtudisc != IP_PMTUDISC_INTERFACE && |
449 | pmtudisc != IP_PMTUDISC_OMIT; |
450 | } |
451 | |
452 | static inline bool ip_sk_use_pmtu(const struct sock *sk) |
453 | { |
454 | return READ_ONCE(inet_sk(sk)->pmtudisc) < IP_PMTUDISC_PROBE; |
455 | } |
456 | |
457 | static inline bool ip_sk_ignore_df(const struct sock *sk) |
458 | { |
459 | u8 pmtudisc = READ_ONCE(inet_sk(sk)->pmtudisc); |
460 | |
461 | return pmtudisc < IP_PMTUDISC_DO || pmtudisc == IP_PMTUDISC_OMIT; |
462 | } |
463 | |
464 | static inline unsigned int ip_dst_mtu_maybe_forward(const struct dst_entry *dst, |
465 | bool forwarding) |
466 | { |
467 | const struct rtable *rt = dst_rtable(dst); |
468 | unsigned int mtu, res; |
469 | struct net *net; |
470 | |
471 | rcu_read_lock(); |
472 | |
473 | net = dev_net_rcu(dev: dst->dev); |
474 | if (READ_ONCE(net->ipv4.sysctl_ip_fwd_use_pmtu) || |
475 | ip_mtu_locked(dst) || |
476 | !forwarding) { |
477 | mtu = rt->rt_pmtu; |
478 | if (mtu && time_before(jiffies, rt->dst.expires)) |
479 | goto out; |
480 | } |
481 | |
482 | /* 'forwarding = true' case should always honour route mtu */ |
483 | mtu = dst_metric_raw(dst, RTAX_MTU); |
484 | if (mtu) |
485 | goto out; |
486 | |
487 | mtu = READ_ONCE(dst->dev->mtu); |
488 | |
489 | if (unlikely(ip_mtu_locked(dst))) { |
490 | if (rt->rt_uses_gateway && mtu > 576) |
491 | mtu = 576; |
492 | } |
493 | |
494 | out: |
495 | mtu = min_t(unsigned int, mtu, IP_MAX_MTU); |
496 | |
497 | res = mtu - lwtunnel_headroom(lwtstate: dst->lwtstate, mtu); |
498 | |
499 | rcu_read_unlock(); |
500 | |
501 | return res; |
502 | } |
503 | |
504 | static inline unsigned int ip_skb_dst_mtu(struct sock *sk, |
505 | const struct sk_buff *skb) |
506 | { |
507 | unsigned int mtu; |
508 | |
509 | if (!sk || !sk_fullsock(sk) || ip_sk_use_pmtu(sk)) { |
510 | bool forwarding = IPCB(skb)->flags & IPSKB_FORWARDED; |
511 | |
512 | return ip_dst_mtu_maybe_forward(dst: skb_dst(skb), forwarding); |
513 | } |
514 | |
515 | mtu = min(READ_ONCE(skb_dst(skb)->dev->mtu), IP_MAX_MTU); |
516 | return mtu - lwtunnel_headroom(lwtstate: skb_dst(skb)->lwtstate, mtu); |
517 | } |
518 | |
519 | struct dst_metrics *ip_fib_metrics_init(struct nlattr *fc_mx, int fc_mx_len, |
520 | struct netlink_ext_ack *extack); |
521 | static inline void ip_fib_metrics_put(struct dst_metrics *fib_metrics) |
522 | { |
523 | if (fib_metrics != &dst_default_metrics && |
524 | refcount_dec_and_test(r: &fib_metrics->refcnt)) |
525 | kfree(objp: fib_metrics); |
526 | } |
527 | |
528 | /* ipv4 and ipv6 both use refcounted metrics if it is not the default */ |
529 | static inline |
530 | void ip_dst_init_metrics(struct dst_entry *dst, struct dst_metrics *fib_metrics) |
531 | { |
532 | dst_init_metrics(dst, src_metrics: fib_metrics->metrics, read_only: true); |
533 | |
534 | if (fib_metrics != &dst_default_metrics) { |
535 | dst->_metrics |= DST_METRICS_REFCOUNTED; |
536 | refcount_inc(r: &fib_metrics->refcnt); |
537 | } |
538 | } |
539 | |
540 | static inline |
541 | void ip_dst_metrics_put(struct dst_entry *dst) |
542 | { |
543 | struct dst_metrics *p = (struct dst_metrics *)DST_METRICS_PTR(dst); |
544 | |
545 | if (p != &dst_default_metrics && refcount_dec_and_test(r: &p->refcnt)) |
546 | kfree(objp: p); |
547 | } |
548 | |
549 | void __ip_select_ident(struct net *net, struct iphdr *iph, int segs); |
550 | |
551 | static inline void ip_select_ident_segs(struct net *net, struct sk_buff *skb, |
552 | struct sock *sk, int segs) |
553 | { |
554 | struct iphdr *iph = ip_hdr(skb); |
555 | |
556 | /* We had many attacks based on IPID, use the private |
557 | * generator as much as we can. |
558 | */ |
559 | if (sk && inet_sk(sk)->inet_daddr) { |
560 | int val; |
561 | |
562 | /* avoid atomic operations for TCP, |
563 | * as we hold socket lock at this point. |
564 | */ |
565 | if (sk_is_tcp(sk)) { |
566 | sock_owned_by_me(sk); |
567 | val = atomic_read(v: &inet_sk(sk)->inet_id); |
568 | atomic_set(v: &inet_sk(sk)->inet_id, i: val + segs); |
569 | } else { |
570 | val = atomic_add_return(i: segs, v: &inet_sk(sk)->inet_id); |
571 | } |
572 | iph->id = htons(val); |
573 | return; |
574 | } |
575 | if ((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) { |
576 | iph->id = 0; |
577 | } else { |
578 | /* Unfortunately we need the big hammer to get a suitable IPID */ |
579 | __ip_select_ident(net, iph, segs); |
580 | } |
581 | } |
582 | |
583 | static inline void ip_select_ident(struct net *net, struct sk_buff *skb, |
584 | struct sock *sk) |
585 | { |
586 | ip_select_ident_segs(net, skb, sk, segs: 1); |
587 | } |
588 | |
589 | static inline __wsum inet_compute_pseudo(struct sk_buff *skb, int proto) |
590 | { |
591 | return csum_tcpudp_nofold(saddr: ip_hdr(skb)->saddr, daddr: ip_hdr(skb)->daddr, |
592 | len: skb->len, proto, sum: 0); |
593 | } |
594 | |
595 | /* copy IPv4 saddr & daddr to flow_keys, possibly using 64bit load/store |
596 | * Equivalent to : flow->v4addrs.src = iph->saddr; |
597 | * flow->v4addrs.dst = iph->daddr; |
598 | */ |
599 | static inline void iph_to_flow_copy_v4addrs(struct flow_keys *flow, |
600 | const struct iphdr *iph) |
601 | { |
602 | BUILD_BUG_ON(offsetof(typeof(flow->addrs), v4addrs.dst) != |
603 | offsetof(typeof(flow->addrs), v4addrs.src) + |
604 | sizeof(flow->addrs.v4addrs.src)); |
605 | memcpy(&flow->addrs.v4addrs, &iph->addrs, sizeof(flow->addrs.v4addrs)); |
606 | flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; |
607 | } |
608 | |
609 | /* |
610 | * Map a multicast IP onto multicast MAC for type ethernet. |
611 | */ |
612 | |
613 | static inline void ip_eth_mc_map(__be32 naddr, char *buf) |
614 | { |
615 | __u32 addr=ntohl(naddr); |
616 | buf[0]=0x01; |
617 | buf[1]=0x00; |
618 | buf[2]=0x5e; |
619 | buf[5]=addr&0xFF; |
620 | addr>>=8; |
621 | buf[4]=addr&0xFF; |
622 | addr>>=8; |
623 | buf[3]=addr&0x7F; |
624 | } |
625 | |
626 | /* |
627 | * Map a multicast IP onto multicast MAC for type IP-over-InfiniBand. |
628 | * Leave P_Key as 0 to be filled in by driver. |
629 | */ |
630 | |
631 | static inline void ip_ib_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) |
632 | { |
633 | __u32 addr; |
634 | unsigned char scope = broadcast[5] & 0xF; |
635 | |
636 | buf[0] = 0; /* Reserved */ |
637 | buf[1] = 0xff; /* Multicast QPN */ |
638 | buf[2] = 0xff; |
639 | buf[3] = 0xff; |
640 | addr = ntohl(naddr); |
641 | buf[4] = 0xff; |
642 | buf[5] = 0x10 | scope; /* scope from broadcast address */ |
643 | buf[6] = 0x40; /* IPv4 signature */ |
644 | buf[7] = 0x1b; |
645 | buf[8] = broadcast[8]; /* P_Key */ |
646 | buf[9] = broadcast[9]; |
647 | buf[10] = 0; |
648 | buf[11] = 0; |
649 | buf[12] = 0; |
650 | buf[13] = 0; |
651 | buf[14] = 0; |
652 | buf[15] = 0; |
653 | buf[19] = addr & 0xff; |
654 | addr >>= 8; |
655 | buf[18] = addr & 0xff; |
656 | addr >>= 8; |
657 | buf[17] = addr & 0xff; |
658 | addr >>= 8; |
659 | buf[16] = addr & 0x0f; |
660 | } |
661 | |
662 | static inline void ip_ipgre_mc_map(__be32 naddr, const unsigned char *broadcast, char *buf) |
663 | { |
664 | if ((broadcast[0] | broadcast[1] | broadcast[2] | broadcast[3]) != 0) |
665 | memcpy(buf, broadcast, 4); |
666 | else |
667 | memcpy(buf, &naddr, sizeof(naddr)); |
668 | } |
669 | |
670 | #if IS_ENABLED(CONFIG_IPV6) |
671 | #include <linux/ipv6.h> |
672 | #endif |
673 | |
674 | static __inline__ void inet_reset_saddr(struct sock *sk) |
675 | { |
676 | inet_sk(sk)->inet_rcv_saddr = inet_sk(sk)->inet_saddr = 0; |
677 | #if IS_ENABLED(CONFIG_IPV6) |
678 | if (sk->sk_family == PF_INET6) { |
679 | struct ipv6_pinfo *np = inet6_sk(sk: sk); |
680 | |
681 | memset(&np->saddr, 0, sizeof(np->saddr)); |
682 | memset(&sk->sk_v6_rcv_saddr, 0, sizeof(sk->sk_v6_rcv_saddr)); |
683 | } |
684 | #endif |
685 | } |
686 | |
687 | #endif |
688 | |
689 | #if IS_MODULE(CONFIG_IPV6) |
690 | #define EXPORT_IPV6_MOD(X) EXPORT_SYMBOL(X) |
691 | #define EXPORT_IPV6_MOD_GPL(X) EXPORT_SYMBOL_GPL(X) |
692 | #else |
693 | #define EXPORT_IPV6_MOD(X) |
694 | #define EXPORT_IPV6_MOD_GPL(X) |
695 | #endif |
696 | |
697 | static inline unsigned int ipv4_addr_hash(__be32 ip) |
698 | { |
699 | return (__force unsigned int) ip; |
700 | } |
701 | |
702 | static inline u32 __ipv4_addr_hash(const __be32 ip, const u32 initval) |
703 | { |
704 | return jhash_1word(a: (__force u32)ip, initval); |
705 | } |
706 | |
707 | static inline u32 ipv4_portaddr_hash(const struct net *net, |
708 | __be32 saddr, |
709 | unsigned int port) |
710 | { |
711 | return jhash_1word(a: (__force u32)saddr, initval: net_hash_mix(net)) ^ port; |
712 | } |
713 | |
714 | bool ip_call_ra_chain(struct sk_buff *skb); |
715 | |
716 | /* |
717 | * Functions provided by ip_fragment.c |
718 | */ |
719 | |
720 | enum ip_defrag_users { |
721 | IP_DEFRAG_LOCAL_DELIVER, |
722 | IP_DEFRAG_CALL_RA_CHAIN, |
723 | IP_DEFRAG_CONNTRACK_IN, |
724 | __IP_DEFRAG_CONNTRACK_IN_END = IP_DEFRAG_CONNTRACK_IN + USHRT_MAX, |
725 | IP_DEFRAG_CONNTRACK_OUT, |
726 | __IP_DEFRAG_CONNTRACK_OUT_END = IP_DEFRAG_CONNTRACK_OUT + USHRT_MAX, |
727 | IP_DEFRAG_CONNTRACK_BRIDGE_IN, |
728 | __IP_DEFRAG_CONNTRACK_BRIDGE_IN = IP_DEFRAG_CONNTRACK_BRIDGE_IN + USHRT_MAX, |
729 | IP_DEFRAG_VS_IN, |
730 | IP_DEFRAG_VS_OUT, |
731 | IP_DEFRAG_VS_FWD, |
732 | IP_DEFRAG_AF_PACKET, |
733 | IP_DEFRAG_MACVLAN, |
734 | }; |
735 | |
736 | /* Return true if the value of 'user' is between 'lower_bond' |
737 | * and 'upper_bond' inclusively. |
738 | */ |
739 | static inline bool ip_defrag_user_in_between(u32 user, |
740 | enum ip_defrag_users lower_bond, |
741 | enum ip_defrag_users upper_bond) |
742 | { |
743 | return user >= lower_bond && user <= upper_bond; |
744 | } |
745 | |
746 | int ip_defrag(struct net *net, struct sk_buff *skb, u32 user); |
747 | #ifdef CONFIG_INET |
748 | struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user); |
749 | #else |
750 | static inline struct sk_buff *ip_check_defrag(struct net *net, struct sk_buff *skb, u32 user) |
751 | { |
752 | return skb; |
753 | } |
754 | #endif |
755 | |
756 | /* |
757 | * Functions provided by ip_forward.c |
758 | */ |
759 | |
760 | int ip_forward(struct sk_buff *skb); |
761 | |
762 | /* |
763 | * Functions provided by ip_options.c |
764 | */ |
765 | |
766 | void ip_options_build(struct sk_buff *skb, struct ip_options *opt, |
767 | __be32 daddr, struct rtable *rt); |
768 | |
769 | int __ip_options_echo(struct net *net, struct ip_options *dopt, |
770 | struct sk_buff *skb, const struct ip_options *sopt); |
771 | static inline int ip_options_echo(struct net *net, struct ip_options *dopt, |
772 | struct sk_buff *skb) |
773 | { |
774 | return __ip_options_echo(net, dopt, skb, sopt: &IPCB(skb)->opt); |
775 | } |
776 | |
777 | void ip_options_fragment(struct sk_buff *skb); |
778 | int __ip_options_compile(struct net *net, struct ip_options *opt, |
779 | struct sk_buff *skb, __be32 *info); |
780 | int ip_options_compile(struct net *net, struct ip_options *opt, |
781 | struct sk_buff *skb); |
782 | int ip_options_get(struct net *net, struct ip_options_rcu **optp, |
783 | sockptr_t data, int optlen); |
784 | void ip_options_undo(struct ip_options *opt); |
785 | void ip_forward_options(struct sk_buff *skb); |
786 | int ip_options_rcv_srr(struct sk_buff *skb, struct net_device *dev); |
787 | |
788 | /* |
789 | * Functions provided by ip_sockglue.c |
790 | */ |
791 | |
792 | void ipv4_pktinfo_prepare(const struct sock *sk, struct sk_buff *skb, bool drop_dst); |
793 | void ip_cmsg_recv_offset(struct msghdr *msg, struct sock *sk, |
794 | struct sk_buff *skb, int tlen, int offset); |
795 | int ip_cmsg_send(struct sock *sk, struct msghdr *msg, |
796 | struct ipcm_cookie *ipc, bool allow_ipv6); |
797 | DECLARE_STATIC_KEY_FALSE(ip4_min_ttl); |
798 | int do_ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, |
799 | unsigned int optlen); |
800 | int ip_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval, |
801 | unsigned int optlen); |
802 | int do_ip_getsockopt(struct sock *sk, int level, int optname, |
803 | sockptr_t optval, sockptr_t optlen); |
804 | int ip_getsockopt(struct sock *sk, int level, int optname, char __user *optval, |
805 | int __user *optlen); |
806 | int ip_ra_control(struct sock *sk, unsigned char on, |
807 | void (*destructor)(struct sock *)); |
808 | |
809 | int ip_recv_error(struct sock *sk, struct msghdr *msg, int len, int *addr_len); |
810 | void ip_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port, |
811 | u32 info, u8 *payload); |
812 | void ip_local_error(struct sock *sk, int err, __be32 daddr, __be16 dport, |
813 | u32 info); |
814 | |
815 | static inline void ip_cmsg_recv(struct msghdr *msg, struct sk_buff *skb) |
816 | { |
817 | ip_cmsg_recv_offset(msg, sk: skb->sk, skb, tlen: 0, offset: 0); |
818 | } |
819 | |
820 | bool icmp_global_allow(struct net *net); |
821 | void icmp_global_consume(struct net *net); |
822 | |
823 | #ifdef CONFIG_PROC_FS |
824 | int ip_misc_proc_init(void); |
825 | #endif |
826 | |
827 | int rtm_getroute_parse_ip_proto(struct nlattr *attr, u8 *ip_proto, u8 family, |
828 | struct netlink_ext_ack *extack); |
829 | |
830 | static inline bool inetdev_valid_mtu(unsigned int mtu) |
831 | { |
832 | return likely(mtu >= IPV4_MIN_MTU); |
833 | } |
834 | |
835 | void ip_sock_set_freebind(struct sock *sk); |
836 | int ip_sock_set_mtu_discover(struct sock *sk, int val); |
837 | void ip_sock_set_pktinfo(struct sock *sk); |
838 | void ip_sock_set_recverr(struct sock *sk); |
839 | void ip_sock_set_tos(struct sock *sk, int val); |
840 | void __ip_sock_set_tos(struct sock *sk, int val); |
841 | |
842 | #endif /* _IP_H */ |
843 |
Definitions
- inet_skb_parm
- ipv4_l3mdev_skb
- ip_hdrlen
- ipcm_cookie
- ipcm_init
- ipcm_init_sk
- inet_sdif
- ip_ra_chain
- ip_fraglist_iter
- ip_fraglist_next
- ip_frag_state
- ip_finish_skb
- ip_sendmsg_scope
- ip_reply_arg
- ip_reply_arg_flowi_flags
- snmp_get_cpu_field
- snmp_get_cpu_field64
- snmp_fold_field64
- inet_get_local_port_range
- inet_is_local_reserved_port
- sysctl_dev_name_is_allowed
- inet_port_requires_bind_service
- ip_is_fragment
- ip_decrease_ttl
- ip4h_dscp
- ip_mtu_locked
- ip_dont_fragment
- ip_sk_accept_pmtu
- ip_sk_use_pmtu
- ip_sk_ignore_df
- ip_dst_mtu_maybe_forward
- ip_skb_dst_mtu
- ip_fib_metrics_put
- ip_dst_init_metrics
- ip_dst_metrics_put
- ip_select_ident_segs
- ip_select_ident
- inet_compute_pseudo
- iph_to_flow_copy_v4addrs
- ip_eth_mc_map
- ip_ib_mc_map
- ip_ipgre_mc_map
- inet_reset_saddr
- ipv4_addr_hash
- __ipv4_addr_hash
- ipv4_portaddr_hash
- ip_defrag_users
- ip_defrag_user_in_between
- ip_options_echo
- ip_cmsg_recv
Improve your Profiling and Debugging skills
Find out more