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