Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * INET An implementation of the TCP/IP protocol suite for the LINUX | |
3 | * operating system. INET is implemented using the BSD Socket | |
4 | * interface as the means of communication with the user level. | |
5 | * | |
6 | * ROUTE - implementation of the IP router. | |
7 | * | |
02c30a84 | 8 | * Authors: Ross Biro |
1da177e4 LT |
9 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
10 | * Alan Cox, <gw4pts@gw4pts.ampr.org> | |
11 | * Linus Torvalds, <Linus.Torvalds@helsinki.fi> | |
12 | * Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru> | |
13 | * | |
14 | * Fixes: | |
15 | * Alan Cox : Verify area fixes. | |
16 | * Alan Cox : cli() protects routing changes | |
17 | * Rui Oliveira : ICMP routing table updates | |
18 | * (rco@di.uminho.pt) Routing table insertion and update | |
19 | * Linus Torvalds : Rewrote bits to be sensible | |
20 | * Alan Cox : Added BSD route gw semantics | |
e905a9ed | 21 | * Alan Cox : Super /proc >4K |
1da177e4 LT |
22 | * Alan Cox : MTU in route table |
23 | * Alan Cox : MSS actually. Also added the window | |
24 | * clamper. | |
25 | * Sam Lantinga : Fixed route matching in rt_del() | |
26 | * Alan Cox : Routing cache support. | |
27 | * Alan Cox : Removed compatibility cruft. | |
28 | * Alan Cox : RTF_REJECT support. | |
29 | * Alan Cox : TCP irtt support. | |
30 | * Jonathan Naylor : Added Metric support. | |
31 | * Miquel van Smoorenburg : BSD API fixes. | |
32 | * Miquel van Smoorenburg : Metrics. | |
33 | * Alan Cox : Use __u32 properly | |
34 | * Alan Cox : Aligned routing errors more closely with BSD | |
35 | * our system is still very different. | |
36 | * Alan Cox : Faster /proc handling | |
37 | * Alexey Kuznetsov : Massive rework to support tree based routing, | |
38 | * routing caches and better behaviour. | |
e905a9ed | 39 | * |
1da177e4 LT |
40 | * Olaf Erb : irtt wasn't being copied right. |
41 | * Bjorn Ekwall : Kerneld route support. | |
42 | * Alan Cox : Multicast fixed (I hope) | |
43 | * Pavel Krauz : Limited broadcast fixed | |
44 | * Mike McLagan : Routing by source | |
45 | * Alexey Kuznetsov : End of old history. Split to fib.c and | |
46 | * route.c and rewritten from scratch. | |
47 | * Andi Kleen : Load-limit warning messages. | |
48 | * Vitaly E. Lavrov : Transparent proxy revived after year coma. | |
49 | * Vitaly E. Lavrov : Race condition in ip_route_input_slow. | |
50 | * Tobias Ringstrom : Uninitialized res.type in ip_route_output_slow. | |
51 | * Vladimir V. Ivanov : IP rule info (flowid) is really useful. | |
52 | * Marc Boucher : routing by fwmark | |
53 | * Robert Olsson : Added rt_cache statistics | |
54 | * Arnaldo C. Melo : Convert proc stuff to seq_file | |
bb1d23b0 | 55 | * Eric Dumazet : hashed spinlocks and rt_check_expire() fixes. |
cef2685e IS |
56 | * Ilia Sotnikov : Ignore TOS on PMTUD and Redirect |
57 | * Ilia Sotnikov : Removed TOS from hash calculations | |
1da177e4 LT |
58 | * |
59 | * This program is free software; you can redistribute it and/or | |
60 | * modify it under the terms of the GNU General Public License | |
61 | * as published by the Free Software Foundation; either version | |
62 | * 2 of the License, or (at your option) any later version. | |
63 | */ | |
64 | ||
afd46503 JP |
65 | #define pr_fmt(fmt) "IPv4: " fmt |
66 | ||
1da177e4 LT |
67 | #include <linux/module.h> |
68 | #include <asm/uaccess.h> | |
1da177e4 LT |
69 | #include <linux/bitops.h> |
70 | #include <linux/types.h> | |
71 | #include <linux/kernel.h> | |
1da177e4 LT |
72 | #include <linux/mm.h> |
73 | #include <linux/string.h> | |
74 | #include <linux/socket.h> | |
75 | #include <linux/sockios.h> | |
76 | #include <linux/errno.h> | |
77 | #include <linux/in.h> | |
78 | #include <linux/inet.h> | |
79 | #include <linux/netdevice.h> | |
80 | #include <linux/proc_fs.h> | |
81 | #include <linux/init.h> | |
82 | #include <linux/skbuff.h> | |
1da177e4 LT |
83 | #include <linux/inetdevice.h> |
84 | #include <linux/igmp.h> | |
85 | #include <linux/pkt_sched.h> | |
86 | #include <linux/mroute.h> | |
87 | #include <linux/netfilter_ipv4.h> | |
88 | #include <linux/random.h> | |
1da177e4 LT |
89 | #include <linux/rcupdate.h> |
90 | #include <linux/times.h> | |
5a0e3ad6 | 91 | #include <linux/slab.h> |
73f156a6 | 92 | #include <linux/jhash.h> |
352e512c | 93 | #include <net/dst.h> |
457c4cbc | 94 | #include <net/net_namespace.h> |
1da177e4 LT |
95 | #include <net/protocol.h> |
96 | #include <net/ip.h> | |
97 | #include <net/route.h> | |
98 | #include <net/inetpeer.h> | |
99 | #include <net/sock.h> | |
100 | #include <net/ip_fib.h> | |
101 | #include <net/arp.h> | |
102 | #include <net/tcp.h> | |
103 | #include <net/icmp.h> | |
104 | #include <net/xfrm.h> | |
8d71740c | 105 | #include <net/netevent.h> |
63f3444f | 106 | #include <net/rtnetlink.h> |
1da177e4 LT |
107 | #ifdef CONFIG_SYSCTL |
108 | #include <linux/sysctl.h> | |
7426a564 | 109 | #include <linux/kmemleak.h> |
1da177e4 | 110 | #endif |
6e5714ea | 111 | #include <net/secure_seq.h> |
1da177e4 | 112 | |
68a5e3dd | 113 | #define RT_FL_TOS(oldflp4) \ |
f61759e6 | 114 | ((oldflp4)->flowi4_tos & (IPTOS_RT_MASK | RTO_ONLINK)) |
1da177e4 | 115 | |
1da177e4 LT |
116 | #define RT_GC_TIMEOUT (300*HZ) |
117 | ||
1da177e4 | 118 | static int ip_rt_max_size; |
817bc4db SH |
119 | static int ip_rt_redirect_number __read_mostly = 9; |
120 | static int ip_rt_redirect_load __read_mostly = HZ / 50; | |
121 | static int ip_rt_redirect_silence __read_mostly = ((HZ / 50) << (9 + 1)); | |
122 | static int ip_rt_error_cost __read_mostly = HZ; | |
123 | static int ip_rt_error_burst __read_mostly = 5 * HZ; | |
817bc4db SH |
124 | static int ip_rt_mtu_expires __read_mostly = 10 * 60 * HZ; |
125 | static int ip_rt_min_pmtu __read_mostly = 512 + 20 + 20; | |
126 | static int ip_rt_min_advmss __read_mostly = 256; | |
9f28a2fc | 127 | |
1da177e4 LT |
128 | /* |
129 | * Interface to generic destination cache. | |
130 | */ | |
131 | ||
132 | static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie); | |
0dbaee3b | 133 | static unsigned int ipv4_default_advmss(const struct dst_entry *dst); |
ebb762f2 | 134 | static unsigned int ipv4_mtu(const struct dst_entry *dst); |
1da177e4 LT |
135 | static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst); |
136 | static void ipv4_link_failure(struct sk_buff *skb); | |
6700c270 DM |
137 | static void ip_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, |
138 | struct sk_buff *skb, u32 mtu); | |
139 | static void ip_do_redirect(struct dst_entry *dst, struct sock *sk, | |
140 | struct sk_buff *skb); | |
caacf05e | 141 | static void ipv4_dst_destroy(struct dst_entry *dst); |
1da177e4 | 142 | |
62fa8a84 DM |
143 | static u32 *ipv4_cow_metrics(struct dst_entry *dst, unsigned long old) |
144 | { | |
31248731 DM |
145 | WARN_ON(1); |
146 | return NULL; | |
62fa8a84 DM |
147 | } |
148 | ||
f894cbf8 DM |
149 | static struct neighbour *ipv4_neigh_lookup(const struct dst_entry *dst, |
150 | struct sk_buff *skb, | |
151 | const void *daddr); | |
d3aaeb38 | 152 | |
1da177e4 LT |
153 | static struct dst_ops ipv4_dst_ops = { |
154 | .family = AF_INET, | |
09640e63 | 155 | .protocol = cpu_to_be16(ETH_P_IP), |
1da177e4 | 156 | .check = ipv4_dst_check, |
0dbaee3b | 157 | .default_advmss = ipv4_default_advmss, |
ebb762f2 | 158 | .mtu = ipv4_mtu, |
62fa8a84 | 159 | .cow_metrics = ipv4_cow_metrics, |
caacf05e | 160 | .destroy = ipv4_dst_destroy, |
1da177e4 LT |
161 | .negative_advice = ipv4_negative_advice, |
162 | .link_failure = ipv4_link_failure, | |
163 | .update_pmtu = ip_rt_update_pmtu, | |
e47a185b | 164 | .redirect = ip_do_redirect, |
1ac06e03 | 165 | .local_out = __ip_local_out, |
d3aaeb38 | 166 | .neigh_lookup = ipv4_neigh_lookup, |
1da177e4 LT |
167 | }; |
168 | ||
169 | #define ECN_OR_COST(class) TC_PRIO_##class | |
170 | ||
4839c52b | 171 | const __u8 ip_tos2prio[16] = { |
1da177e4 | 172 | TC_PRIO_BESTEFFORT, |
4a2b9c37 | 173 | ECN_OR_COST(BESTEFFORT), |
1da177e4 LT |
174 | TC_PRIO_BESTEFFORT, |
175 | ECN_OR_COST(BESTEFFORT), | |
176 | TC_PRIO_BULK, | |
177 | ECN_OR_COST(BULK), | |
178 | TC_PRIO_BULK, | |
179 | ECN_OR_COST(BULK), | |
180 | TC_PRIO_INTERACTIVE, | |
181 | ECN_OR_COST(INTERACTIVE), | |
182 | TC_PRIO_INTERACTIVE, | |
183 | ECN_OR_COST(INTERACTIVE), | |
184 | TC_PRIO_INTERACTIVE_BULK, | |
185 | ECN_OR_COST(INTERACTIVE_BULK), | |
186 | TC_PRIO_INTERACTIVE_BULK, | |
187 | ECN_OR_COST(INTERACTIVE_BULK) | |
188 | }; | |
d4a96865 | 189 | EXPORT_SYMBOL(ip_tos2prio); |
1da177e4 | 190 | |
2f970d83 | 191 | static DEFINE_PER_CPU(struct rt_cache_stat, rt_cache_stat); |
3ed66e91 | 192 | #define RT_CACHE_STAT_INC(field) raw_cpu_inc(rt_cache_stat.field) |
1da177e4 | 193 | |
1da177e4 | 194 | #ifdef CONFIG_PROC_FS |
1da177e4 LT |
195 | static void *rt_cache_seq_start(struct seq_file *seq, loff_t *pos) |
196 | { | |
29e75252 | 197 | if (*pos) |
89aef892 | 198 | return NULL; |
29e75252 | 199 | return SEQ_START_TOKEN; |
1da177e4 LT |
200 | } |
201 | ||
202 | static void *rt_cache_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
203 | { | |
1da177e4 | 204 | ++*pos; |
89aef892 | 205 | return NULL; |
1da177e4 LT |
206 | } |
207 | ||
208 | static void rt_cache_seq_stop(struct seq_file *seq, void *v) | |
209 | { | |
1da177e4 LT |
210 | } |
211 | ||
212 | static int rt_cache_seq_show(struct seq_file *seq, void *v) | |
213 | { | |
214 | if (v == SEQ_START_TOKEN) | |
215 | seq_printf(seq, "%-127s\n", | |
216 | "Iface\tDestination\tGateway \tFlags\t\tRefCnt\tUse\t" | |
217 | "Metric\tSource\t\tMTU\tWindow\tIRTT\tTOS\tHHRef\t" | |
218 | "HHUptod\tSpecDst"); | |
e905a9ed | 219 | return 0; |
1da177e4 LT |
220 | } |
221 | ||
f690808e | 222 | static const struct seq_operations rt_cache_seq_ops = { |
1da177e4 LT |
223 | .start = rt_cache_seq_start, |
224 | .next = rt_cache_seq_next, | |
225 | .stop = rt_cache_seq_stop, | |
226 | .show = rt_cache_seq_show, | |
227 | }; | |
228 | ||
229 | static int rt_cache_seq_open(struct inode *inode, struct file *file) | |
230 | { | |
89aef892 | 231 | return seq_open(file, &rt_cache_seq_ops); |
1da177e4 LT |
232 | } |
233 | ||
9a32144e | 234 | static const struct file_operations rt_cache_seq_fops = { |
1da177e4 LT |
235 | .owner = THIS_MODULE, |
236 | .open = rt_cache_seq_open, | |
237 | .read = seq_read, | |
238 | .llseek = seq_lseek, | |
89aef892 | 239 | .release = seq_release, |
1da177e4 LT |
240 | }; |
241 | ||
242 | ||
243 | static void *rt_cpu_seq_start(struct seq_file *seq, loff_t *pos) | |
244 | { | |
245 | int cpu; | |
246 | ||
247 | if (*pos == 0) | |
248 | return SEQ_START_TOKEN; | |
249 | ||
0f23174a | 250 | for (cpu = *pos-1; cpu < nr_cpu_ids; ++cpu) { |
1da177e4 LT |
251 | if (!cpu_possible(cpu)) |
252 | continue; | |
253 | *pos = cpu+1; | |
2f970d83 | 254 | return &per_cpu(rt_cache_stat, cpu); |
1da177e4 LT |
255 | } |
256 | return NULL; | |
257 | } | |
258 | ||
259 | static void *rt_cpu_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
260 | { | |
261 | int cpu; | |
262 | ||
0f23174a | 263 | for (cpu = *pos; cpu < nr_cpu_ids; ++cpu) { |
1da177e4 LT |
264 | if (!cpu_possible(cpu)) |
265 | continue; | |
266 | *pos = cpu+1; | |
2f970d83 | 267 | return &per_cpu(rt_cache_stat, cpu); |
1da177e4 LT |
268 | } |
269 | return NULL; | |
e905a9ed | 270 | |
1da177e4 LT |
271 | } |
272 | ||
273 | static void rt_cpu_seq_stop(struct seq_file *seq, void *v) | |
274 | { | |
275 | ||
276 | } | |
277 | ||
278 | static int rt_cpu_seq_show(struct seq_file *seq, void *v) | |
279 | { | |
280 | struct rt_cache_stat *st = v; | |
281 | ||
282 | if (v == SEQ_START_TOKEN) { | |
5bec0039 | 283 | seq_printf(seq, "entries in_hit in_slow_tot in_slow_mc in_no_route in_brd in_martian_dst in_martian_src out_hit out_slow_tot out_slow_mc gc_total gc_ignored gc_goal_miss gc_dst_overflow in_hlist_search out_hlist_search\n"); |
1da177e4 LT |
284 | return 0; |
285 | } | |
e905a9ed | 286 | |
1da177e4 LT |
287 | seq_printf(seq,"%08x %08x %08x %08x %08x %08x %08x %08x " |
288 | " %08x %08x %08x %08x %08x %08x %08x %08x %08x \n", | |
fc66f95c | 289 | dst_entries_get_slow(&ipv4_dst_ops), |
0baf2b35 | 290 | 0, /* st->in_hit */ |
1da177e4 LT |
291 | st->in_slow_tot, |
292 | st->in_slow_mc, | |
293 | st->in_no_route, | |
294 | st->in_brd, | |
295 | st->in_martian_dst, | |
296 | st->in_martian_src, | |
297 | ||
0baf2b35 | 298 | 0, /* st->out_hit */ |
1da177e4 | 299 | st->out_slow_tot, |
e905a9ed | 300 | st->out_slow_mc, |
1da177e4 | 301 | |
0baf2b35 ED |
302 | 0, /* st->gc_total */ |
303 | 0, /* st->gc_ignored */ | |
304 | 0, /* st->gc_goal_miss */ | |
305 | 0, /* st->gc_dst_overflow */ | |
306 | 0, /* st->in_hlist_search */ | |
307 | 0 /* st->out_hlist_search */ | |
1da177e4 LT |
308 | ); |
309 | return 0; | |
310 | } | |
311 | ||
f690808e | 312 | static const struct seq_operations rt_cpu_seq_ops = { |
1da177e4 LT |
313 | .start = rt_cpu_seq_start, |
314 | .next = rt_cpu_seq_next, | |
315 | .stop = rt_cpu_seq_stop, | |
316 | .show = rt_cpu_seq_show, | |
317 | }; | |
318 | ||
319 | ||
320 | static int rt_cpu_seq_open(struct inode *inode, struct file *file) | |
321 | { | |
322 | return seq_open(file, &rt_cpu_seq_ops); | |
323 | } | |
324 | ||
9a32144e | 325 | static const struct file_operations rt_cpu_seq_fops = { |
1da177e4 LT |
326 | .owner = THIS_MODULE, |
327 | .open = rt_cpu_seq_open, | |
328 | .read = seq_read, | |
329 | .llseek = seq_lseek, | |
330 | .release = seq_release, | |
331 | }; | |
332 | ||
c7066f70 | 333 | #ifdef CONFIG_IP_ROUTE_CLASSID |
a661c419 | 334 | static int rt_acct_proc_show(struct seq_file *m, void *v) |
78c686e9 | 335 | { |
a661c419 AD |
336 | struct ip_rt_acct *dst, *src; |
337 | unsigned int i, j; | |
338 | ||
339 | dst = kcalloc(256, sizeof(struct ip_rt_acct), GFP_KERNEL); | |
340 | if (!dst) | |
341 | return -ENOMEM; | |
342 | ||
343 | for_each_possible_cpu(i) { | |
344 | src = (struct ip_rt_acct *)per_cpu_ptr(ip_rt_acct, i); | |
345 | for (j = 0; j < 256; j++) { | |
346 | dst[j].o_bytes += src[j].o_bytes; | |
347 | dst[j].o_packets += src[j].o_packets; | |
348 | dst[j].i_bytes += src[j].i_bytes; | |
349 | dst[j].i_packets += src[j].i_packets; | |
350 | } | |
78c686e9 PE |
351 | } |
352 | ||
a661c419 AD |
353 | seq_write(m, dst, 256 * sizeof(struct ip_rt_acct)); |
354 | kfree(dst); | |
355 | return 0; | |
356 | } | |
78c686e9 | 357 | |
a661c419 AD |
358 | static int rt_acct_proc_open(struct inode *inode, struct file *file) |
359 | { | |
360 | return single_open(file, rt_acct_proc_show, NULL); | |
78c686e9 | 361 | } |
a661c419 AD |
362 | |
363 | static const struct file_operations rt_acct_proc_fops = { | |
364 | .owner = THIS_MODULE, | |
365 | .open = rt_acct_proc_open, | |
366 | .read = seq_read, | |
367 | .llseek = seq_lseek, | |
368 | .release = single_release, | |
369 | }; | |
78c686e9 | 370 | #endif |
107f1634 | 371 | |
73b38711 | 372 | static int __net_init ip_rt_do_proc_init(struct net *net) |
107f1634 PE |
373 | { |
374 | struct proc_dir_entry *pde; | |
375 | ||
d4beaa66 G |
376 | pde = proc_create("rt_cache", S_IRUGO, net->proc_net, |
377 | &rt_cache_seq_fops); | |
107f1634 PE |
378 | if (!pde) |
379 | goto err1; | |
380 | ||
77020720 WC |
381 | pde = proc_create("rt_cache", S_IRUGO, |
382 | net->proc_net_stat, &rt_cpu_seq_fops); | |
107f1634 PE |
383 | if (!pde) |
384 | goto err2; | |
385 | ||
c7066f70 | 386 | #ifdef CONFIG_IP_ROUTE_CLASSID |
a661c419 | 387 | pde = proc_create("rt_acct", 0, net->proc_net, &rt_acct_proc_fops); |
107f1634 PE |
388 | if (!pde) |
389 | goto err3; | |
390 | #endif | |
391 | return 0; | |
392 | ||
c7066f70 | 393 | #ifdef CONFIG_IP_ROUTE_CLASSID |
107f1634 PE |
394 | err3: |
395 | remove_proc_entry("rt_cache", net->proc_net_stat); | |
396 | #endif | |
397 | err2: | |
398 | remove_proc_entry("rt_cache", net->proc_net); | |
399 | err1: | |
400 | return -ENOMEM; | |
401 | } | |
73b38711 DL |
402 | |
403 | static void __net_exit ip_rt_do_proc_exit(struct net *net) | |
404 | { | |
405 | remove_proc_entry("rt_cache", net->proc_net_stat); | |
406 | remove_proc_entry("rt_cache", net->proc_net); | |
c7066f70 | 407 | #ifdef CONFIG_IP_ROUTE_CLASSID |
73b38711 | 408 | remove_proc_entry("rt_acct", net->proc_net); |
0a931acf | 409 | #endif |
73b38711 DL |
410 | } |
411 | ||
412 | static struct pernet_operations ip_rt_proc_ops __net_initdata = { | |
413 | .init = ip_rt_do_proc_init, | |
414 | .exit = ip_rt_do_proc_exit, | |
415 | }; | |
416 | ||
417 | static int __init ip_rt_proc_init(void) | |
418 | { | |
419 | return register_pernet_subsys(&ip_rt_proc_ops); | |
420 | } | |
421 | ||
107f1634 | 422 | #else |
73b38711 | 423 | static inline int ip_rt_proc_init(void) |
107f1634 PE |
424 | { |
425 | return 0; | |
426 | } | |
1da177e4 | 427 | #endif /* CONFIG_PROC_FS */ |
e905a9ed | 428 | |
4331debc | 429 | static inline bool rt_is_expired(const struct rtable *rth) |
e84f84f2 | 430 | { |
ca4c3fc2 | 431 | return rth->rt_genid != rt_genid_ipv4(dev_net(rth->dst.dev)); |
e84f84f2 DL |
432 | } |
433 | ||
4ccfe6d4 | 434 | void rt_cache_flush(struct net *net) |
1da177e4 | 435 | { |
ca4c3fc2 | 436 | rt_genid_bump_ipv4(net); |
98376387 ED |
437 | } |
438 | ||
f894cbf8 DM |
439 | static struct neighbour *ipv4_neigh_lookup(const struct dst_entry *dst, |
440 | struct sk_buff *skb, | |
441 | const void *daddr) | |
3769cffb | 442 | { |
d3aaeb38 DM |
443 | struct net_device *dev = dst->dev; |
444 | const __be32 *pkey = daddr; | |
39232973 | 445 | const struct rtable *rt; |
3769cffb DM |
446 | struct neighbour *n; |
447 | ||
39232973 | 448 | rt = (const struct rtable *) dst; |
a263b309 | 449 | if (rt->rt_gateway) |
39232973 | 450 | pkey = (const __be32 *) &rt->rt_gateway; |
f894cbf8 DM |
451 | else if (skb) |
452 | pkey = &ip_hdr(skb)->daddr; | |
d3aaeb38 | 453 | |
80703d26 | 454 | n = __ipv4_neigh_lookup(dev, *(__force u32 *)pkey); |
d3aaeb38 DM |
455 | if (n) |
456 | return n; | |
32092ecf | 457 | return neigh_create(&arp_tbl, pkey, dev); |
d3aaeb38 DM |
458 | } |
459 | ||
73f156a6 ED |
460 | atomic_t *ip_idents __read_mostly; |
461 | EXPORT_SYMBOL(ip_idents); | |
1da177e4 | 462 | |
73f156a6 | 463 | void __ip_select_ident(struct iphdr *iph, int segs) |
1da177e4 | 464 | { |
73f156a6 ED |
465 | static u32 ip_idents_hashrnd __read_mostly; |
466 | u32 hash, id; | |
1da177e4 | 467 | |
73f156a6 | 468 | net_get_random_once(&ip_idents_hashrnd, sizeof(ip_idents_hashrnd)); |
1da177e4 | 469 | |
73f156a6 ED |
470 | hash = jhash_1word((__force u32)iph->daddr, ip_idents_hashrnd); |
471 | id = ip_idents_reserve(hash, segs); | |
472 | iph->id = htons(id); | |
1da177e4 | 473 | } |
4bc2f18b | 474 | EXPORT_SYMBOL(__ip_select_ident); |
1da177e4 | 475 | |
5abf7f7e | 476 | static void __build_flow_key(struct flowi4 *fl4, const struct sock *sk, |
4895c771 DM |
477 | const struct iphdr *iph, |
478 | int oif, u8 tos, | |
479 | u8 prot, u32 mark, int flow_flags) | |
480 | { | |
481 | if (sk) { | |
482 | const struct inet_sock *inet = inet_sk(sk); | |
483 | ||
484 | oif = sk->sk_bound_dev_if; | |
485 | mark = sk->sk_mark; | |
486 | tos = RT_CONN_FLAGS(sk); | |
487 | prot = inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol; | |
488 | } | |
489 | flowi4_init_output(fl4, oif, mark, tos, | |
490 | RT_SCOPE_UNIVERSE, prot, | |
491 | flow_flags, | |
492 | iph->daddr, iph->saddr, 0, 0); | |
493 | } | |
494 | ||
5abf7f7e ED |
495 | static void build_skb_flow_key(struct flowi4 *fl4, const struct sk_buff *skb, |
496 | const struct sock *sk) | |
4895c771 DM |
497 | { |
498 | const struct iphdr *iph = ip_hdr(skb); | |
499 | int oif = skb->dev->ifindex; | |
500 | u8 tos = RT_TOS(iph->tos); | |
501 | u8 prot = iph->protocol; | |
502 | u32 mark = skb->mark; | |
503 | ||
504 | __build_flow_key(fl4, sk, iph, oif, tos, prot, mark, 0); | |
505 | } | |
506 | ||
5abf7f7e | 507 | static void build_sk_flow_key(struct flowi4 *fl4, const struct sock *sk) |
4895c771 DM |
508 | { |
509 | const struct inet_sock *inet = inet_sk(sk); | |
5abf7f7e | 510 | const struct ip_options_rcu *inet_opt; |
4895c771 DM |
511 | __be32 daddr = inet->inet_daddr; |
512 | ||
513 | rcu_read_lock(); | |
514 | inet_opt = rcu_dereference(inet->inet_opt); | |
515 | if (inet_opt && inet_opt->opt.srr) | |
516 | daddr = inet_opt->opt.faddr; | |
517 | flowi4_init_output(fl4, sk->sk_bound_dev_if, sk->sk_mark, | |
518 | RT_CONN_FLAGS(sk), RT_SCOPE_UNIVERSE, | |
519 | inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol, | |
520 | inet_sk_flowi_flags(sk), | |
521 | daddr, inet->inet_saddr, 0, 0); | |
522 | rcu_read_unlock(); | |
523 | } | |
524 | ||
5abf7f7e ED |
525 | static void ip_rt_build_flow_key(struct flowi4 *fl4, const struct sock *sk, |
526 | const struct sk_buff *skb) | |
4895c771 DM |
527 | { |
528 | if (skb) | |
529 | build_skb_flow_key(fl4, skb, sk); | |
530 | else | |
531 | build_sk_flow_key(fl4, sk); | |
532 | } | |
533 | ||
c5038a83 DM |
534 | static inline void rt_free(struct rtable *rt) |
535 | { | |
536 | call_rcu(&rt->dst.rcu_head, dst_rcu_free); | |
537 | } | |
538 | ||
539 | static DEFINE_SPINLOCK(fnhe_lock); | |
4895c771 | 540 | |
2ffae99d TT |
541 | static void fnhe_flush_routes(struct fib_nh_exception *fnhe) |
542 | { | |
543 | struct rtable *rt; | |
544 | ||
545 | rt = rcu_dereference(fnhe->fnhe_rth_input); | |
546 | if (rt) { | |
547 | RCU_INIT_POINTER(fnhe->fnhe_rth_input, NULL); | |
548 | rt_free(rt); | |
549 | } | |
550 | rt = rcu_dereference(fnhe->fnhe_rth_output); | |
551 | if (rt) { | |
552 | RCU_INIT_POINTER(fnhe->fnhe_rth_output, NULL); | |
553 | rt_free(rt); | |
554 | } | |
555 | } | |
556 | ||
aee06da6 | 557 | static struct fib_nh_exception *fnhe_oldest(struct fnhe_hash_bucket *hash) |
4895c771 DM |
558 | { |
559 | struct fib_nh_exception *fnhe, *oldest; | |
560 | ||
561 | oldest = rcu_dereference(hash->chain); | |
562 | for (fnhe = rcu_dereference(oldest->fnhe_next); fnhe; | |
563 | fnhe = rcu_dereference(fnhe->fnhe_next)) { | |
564 | if (time_before(fnhe->fnhe_stamp, oldest->fnhe_stamp)) | |
565 | oldest = fnhe; | |
566 | } | |
2ffae99d | 567 | fnhe_flush_routes(oldest); |
4895c771 DM |
568 | return oldest; |
569 | } | |
570 | ||
d3a25c98 DM |
571 | static inline u32 fnhe_hashfun(__be32 daddr) |
572 | { | |
573 | u32 hval; | |
574 | ||
575 | hval = (__force u32) daddr; | |
576 | hval ^= (hval >> 11) ^ (hval >> 22); | |
577 | ||
578 | return hval & (FNHE_HASH_SIZE - 1); | |
579 | } | |
580 | ||
387aa65a TT |
581 | static void fill_route_from_fnhe(struct rtable *rt, struct fib_nh_exception *fnhe) |
582 | { | |
583 | rt->rt_pmtu = fnhe->fnhe_pmtu; | |
584 | rt->dst.expires = fnhe->fnhe_expires; | |
585 | ||
586 | if (fnhe->fnhe_gw) { | |
587 | rt->rt_flags |= RTCF_REDIRECTED; | |
588 | rt->rt_gateway = fnhe->fnhe_gw; | |
589 | rt->rt_uses_gateway = 1; | |
590 | } | |
591 | } | |
592 | ||
aee06da6 JA |
593 | static void update_or_create_fnhe(struct fib_nh *nh, __be32 daddr, __be32 gw, |
594 | u32 pmtu, unsigned long expires) | |
4895c771 | 595 | { |
aee06da6 | 596 | struct fnhe_hash_bucket *hash; |
4895c771 | 597 | struct fib_nh_exception *fnhe; |
387aa65a TT |
598 | struct rtable *rt; |
599 | unsigned int i; | |
4895c771 | 600 | int depth; |
aee06da6 JA |
601 | u32 hval = fnhe_hashfun(daddr); |
602 | ||
c5038a83 | 603 | spin_lock_bh(&fnhe_lock); |
4895c771 | 604 | |
aee06da6 | 605 | hash = nh->nh_exceptions; |
4895c771 | 606 | if (!hash) { |
aee06da6 | 607 | hash = kzalloc(FNHE_HASH_SIZE * sizeof(*hash), GFP_ATOMIC); |
4895c771 | 608 | if (!hash) |
aee06da6 JA |
609 | goto out_unlock; |
610 | nh->nh_exceptions = hash; | |
4895c771 DM |
611 | } |
612 | ||
4895c771 DM |
613 | hash += hval; |
614 | ||
615 | depth = 0; | |
616 | for (fnhe = rcu_dereference(hash->chain); fnhe; | |
617 | fnhe = rcu_dereference(fnhe->fnhe_next)) { | |
618 | if (fnhe->fnhe_daddr == daddr) | |
aee06da6 | 619 | break; |
4895c771 DM |
620 | depth++; |
621 | } | |
622 | ||
aee06da6 JA |
623 | if (fnhe) { |
624 | if (gw) | |
625 | fnhe->fnhe_gw = gw; | |
626 | if (pmtu) { | |
627 | fnhe->fnhe_pmtu = pmtu; | |
387aa65a | 628 | fnhe->fnhe_expires = max(1UL, expires); |
aee06da6 | 629 | } |
387aa65a | 630 | /* Update all cached dsts too */ |
2ffae99d TT |
631 | rt = rcu_dereference(fnhe->fnhe_rth_input); |
632 | if (rt) | |
633 | fill_route_from_fnhe(rt, fnhe); | |
634 | rt = rcu_dereference(fnhe->fnhe_rth_output); | |
387aa65a TT |
635 | if (rt) |
636 | fill_route_from_fnhe(rt, fnhe); | |
aee06da6 JA |
637 | } else { |
638 | if (depth > FNHE_RECLAIM_DEPTH) | |
639 | fnhe = fnhe_oldest(hash); | |
640 | else { | |
641 | fnhe = kzalloc(sizeof(*fnhe), GFP_ATOMIC); | |
642 | if (!fnhe) | |
643 | goto out_unlock; | |
644 | ||
645 | fnhe->fnhe_next = hash->chain; | |
646 | rcu_assign_pointer(hash->chain, fnhe); | |
647 | } | |
5aad1de5 | 648 | fnhe->fnhe_genid = fnhe_genid(dev_net(nh->nh_dev)); |
aee06da6 JA |
649 | fnhe->fnhe_daddr = daddr; |
650 | fnhe->fnhe_gw = gw; | |
651 | fnhe->fnhe_pmtu = pmtu; | |
652 | fnhe->fnhe_expires = expires; | |
387aa65a TT |
653 | |
654 | /* Exception created; mark the cached routes for the nexthop | |
655 | * stale, so anyone caching it rechecks if this exception | |
656 | * applies to them. | |
657 | */ | |
2ffae99d TT |
658 | rt = rcu_dereference(nh->nh_rth_input); |
659 | if (rt) | |
660 | rt->dst.obsolete = DST_OBSOLETE_KILL; | |
661 | ||
387aa65a TT |
662 | for_each_possible_cpu(i) { |
663 | struct rtable __rcu **prt; | |
664 | prt = per_cpu_ptr(nh->nh_pcpu_rth_output, i); | |
665 | rt = rcu_dereference(*prt); | |
666 | if (rt) | |
667 | rt->dst.obsolete = DST_OBSOLETE_KILL; | |
668 | } | |
4895c771 | 669 | } |
4895c771 | 670 | |
4895c771 | 671 | fnhe->fnhe_stamp = jiffies; |
aee06da6 JA |
672 | |
673 | out_unlock: | |
c5038a83 | 674 | spin_unlock_bh(&fnhe_lock); |
4895c771 DM |
675 | } |
676 | ||
ceb33206 DM |
677 | static void __ip_do_redirect(struct rtable *rt, struct sk_buff *skb, struct flowi4 *fl4, |
678 | bool kill_route) | |
1da177e4 | 679 | { |
e47a185b | 680 | __be32 new_gw = icmp_hdr(skb)->un.gateway; |
94206125 | 681 | __be32 old_gw = ip_hdr(skb)->saddr; |
e47a185b | 682 | struct net_device *dev = skb->dev; |
e47a185b | 683 | struct in_device *in_dev; |
4895c771 | 684 | struct fib_result res; |
e47a185b | 685 | struct neighbour *n; |
317805b8 | 686 | struct net *net; |
1da177e4 | 687 | |
94206125 DM |
688 | switch (icmp_hdr(skb)->code & 7) { |
689 | case ICMP_REDIR_NET: | |
690 | case ICMP_REDIR_NETTOS: | |
691 | case ICMP_REDIR_HOST: | |
692 | case ICMP_REDIR_HOSTTOS: | |
693 | break; | |
694 | ||
695 | default: | |
696 | return; | |
697 | } | |
698 | ||
e47a185b DM |
699 | if (rt->rt_gateway != old_gw) |
700 | return; | |
701 | ||
702 | in_dev = __in_dev_get_rcu(dev); | |
703 | if (!in_dev) | |
704 | return; | |
705 | ||
c346dca1 | 706 | net = dev_net(dev); |
9d4fb27d JP |
707 | if (new_gw == old_gw || !IN_DEV_RX_REDIRECTS(in_dev) || |
708 | ipv4_is_multicast(new_gw) || ipv4_is_lbcast(new_gw) || | |
709 | ipv4_is_zeronet(new_gw)) | |
1da177e4 LT |
710 | goto reject_redirect; |
711 | ||
712 | if (!IN_DEV_SHARED_MEDIA(in_dev)) { | |
713 | if (!inet_addr_onlink(in_dev, new_gw, old_gw)) | |
714 | goto reject_redirect; | |
715 | if (IN_DEV_SEC_REDIRECTS(in_dev) && ip_fib_check_default(new_gw, dev)) | |
716 | goto reject_redirect; | |
717 | } else { | |
317805b8 | 718 | if (inet_addr_type(net, new_gw) != RTN_UNICAST) |
1da177e4 LT |
719 | goto reject_redirect; |
720 | } | |
721 | ||
4895c771 | 722 | n = ipv4_neigh_lookup(&rt->dst, NULL, &new_gw); |
e47a185b DM |
723 | if (n) { |
724 | if (!(n->nud_state & NUD_VALID)) { | |
725 | neigh_event_send(n, NULL); | |
726 | } else { | |
4895c771 DM |
727 | if (fib_lookup(net, fl4, &res) == 0) { |
728 | struct fib_nh *nh = &FIB_RES_NH(res); | |
4895c771 | 729 | |
aee06da6 JA |
730 | update_or_create_fnhe(nh, fl4->daddr, new_gw, |
731 | 0, 0); | |
4895c771 | 732 | } |
ceb33206 DM |
733 | if (kill_route) |
734 | rt->dst.obsolete = DST_OBSOLETE_KILL; | |
e47a185b DM |
735 | call_netevent_notifiers(NETEVENT_NEIGH_UPDATE, n); |
736 | } | |
737 | neigh_release(n); | |
738 | } | |
739 | return; | |
740 | ||
741 | reject_redirect: | |
742 | #ifdef CONFIG_IP_ROUTE_VERBOSE | |
99ee038d DM |
743 | if (IN_DEV_LOG_MARTIANS(in_dev)) { |
744 | const struct iphdr *iph = (const struct iphdr *) skb->data; | |
745 | __be32 daddr = iph->daddr; | |
746 | __be32 saddr = iph->saddr; | |
747 | ||
e47a185b DM |
748 | net_info_ratelimited("Redirect from %pI4 on %s about %pI4 ignored\n" |
749 | " Advised path = %pI4 -> %pI4\n", | |
750 | &old_gw, dev->name, &new_gw, | |
751 | &saddr, &daddr); | |
99ee038d | 752 | } |
e47a185b DM |
753 | #endif |
754 | ; | |
755 | } | |
756 | ||
4895c771 DM |
757 | static void ip_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_buff *skb) |
758 | { | |
759 | struct rtable *rt; | |
760 | struct flowi4 fl4; | |
f96ef988 MK |
761 | const struct iphdr *iph = (const struct iphdr *) skb->data; |
762 | int oif = skb->dev->ifindex; | |
763 | u8 tos = RT_TOS(iph->tos); | |
764 | u8 prot = iph->protocol; | |
765 | u32 mark = skb->mark; | |
4895c771 DM |
766 | |
767 | rt = (struct rtable *) dst; | |
768 | ||
f96ef988 | 769 | __build_flow_key(&fl4, sk, iph, oif, tos, prot, mark, 0); |
ceb33206 | 770 | __ip_do_redirect(rt, skb, &fl4, true); |
4895c771 DM |
771 | } |
772 | ||
1da177e4 LT |
773 | static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst) |
774 | { | |
ee6b9673 | 775 | struct rtable *rt = (struct rtable *)dst; |
1da177e4 LT |
776 | struct dst_entry *ret = dst; |
777 | ||
778 | if (rt) { | |
d11a4dc1 | 779 | if (dst->obsolete > 0) { |
1da177e4 LT |
780 | ip_rt_put(rt); |
781 | ret = NULL; | |
5943634f DM |
782 | } else if ((rt->rt_flags & RTCF_REDIRECTED) || |
783 | rt->dst.expires) { | |
89aef892 | 784 | ip_rt_put(rt); |
1da177e4 LT |
785 | ret = NULL; |
786 | } | |
787 | } | |
788 | return ret; | |
789 | } | |
790 | ||
791 | /* | |
792 | * Algorithm: | |
793 | * 1. The first ip_rt_redirect_number redirects are sent | |
794 | * with exponential backoff, then we stop sending them at all, | |
795 | * assuming that the host ignores our redirects. | |
796 | * 2. If we did not see packets requiring redirects | |
797 | * during ip_rt_redirect_silence, we assume that the host | |
798 | * forgot redirected route and start to send redirects again. | |
799 | * | |
800 | * This algorithm is much cheaper and more intelligent than dumb load limiting | |
801 | * in icmp.c. | |
802 | * | |
803 | * NOTE. Do not forget to inhibit load limiting for redirects (redundant) | |
804 | * and "frag. need" (breaks PMTU discovery) in icmp.c. | |
805 | */ | |
806 | ||
807 | void ip_rt_send_redirect(struct sk_buff *skb) | |
808 | { | |
511c3f92 | 809 | struct rtable *rt = skb_rtable(skb); |
30038fc6 | 810 | struct in_device *in_dev; |
92d86829 | 811 | struct inet_peer *peer; |
1d861aa4 | 812 | struct net *net; |
30038fc6 | 813 | int log_martians; |
1da177e4 | 814 | |
30038fc6 | 815 | rcu_read_lock(); |
d8d1f30b | 816 | in_dev = __in_dev_get_rcu(rt->dst.dev); |
30038fc6 ED |
817 | if (!in_dev || !IN_DEV_TX_REDIRECTS(in_dev)) { |
818 | rcu_read_unlock(); | |
1da177e4 | 819 | return; |
30038fc6 ED |
820 | } |
821 | log_martians = IN_DEV_LOG_MARTIANS(in_dev); | |
822 | rcu_read_unlock(); | |
1da177e4 | 823 | |
1d861aa4 DM |
824 | net = dev_net(rt->dst.dev); |
825 | peer = inet_getpeer_v4(net->ipv4.peers, ip_hdr(skb)->saddr, 1); | |
92d86829 | 826 | if (!peer) { |
e81da0e1 JA |
827 | icmp_send(skb, ICMP_REDIRECT, ICMP_REDIR_HOST, |
828 | rt_nexthop(rt, ip_hdr(skb)->daddr)); | |
92d86829 DM |
829 | return; |
830 | } | |
831 | ||
1da177e4 LT |
832 | /* No redirected packets during ip_rt_redirect_silence; |
833 | * reset the algorithm. | |
834 | */ | |
92d86829 DM |
835 | if (time_after(jiffies, peer->rate_last + ip_rt_redirect_silence)) |
836 | peer->rate_tokens = 0; | |
1da177e4 LT |
837 | |
838 | /* Too many ignored redirects; do not send anything | |
d8d1f30b | 839 | * set dst.rate_last to the last seen redirected packet. |
1da177e4 | 840 | */ |
92d86829 DM |
841 | if (peer->rate_tokens >= ip_rt_redirect_number) { |
842 | peer->rate_last = jiffies; | |
1d861aa4 | 843 | goto out_put_peer; |
1da177e4 LT |
844 | } |
845 | ||
846 | /* Check for load limit; set rate_last to the latest sent | |
847 | * redirect. | |
848 | */ | |
92d86829 | 849 | if (peer->rate_tokens == 0 || |
14fb8a76 | 850 | time_after(jiffies, |
92d86829 DM |
851 | (peer->rate_last + |
852 | (ip_rt_redirect_load << peer->rate_tokens)))) { | |
e81da0e1 JA |
853 | __be32 gw = rt_nexthop(rt, ip_hdr(skb)->daddr); |
854 | ||
855 | icmp_send(skb, ICMP_REDIRECT, ICMP_REDIR_HOST, gw); | |
92d86829 DM |
856 | peer->rate_last = jiffies; |
857 | ++peer->rate_tokens; | |
1da177e4 | 858 | #ifdef CONFIG_IP_ROUTE_VERBOSE |
30038fc6 | 859 | if (log_martians && |
e87cc472 JP |
860 | peer->rate_tokens == ip_rt_redirect_number) |
861 | net_warn_ratelimited("host %pI4/if%d ignores redirects for %pI4 to %pI4\n", | |
92101b3b | 862 | &ip_hdr(skb)->saddr, inet_iif(skb), |
e81da0e1 | 863 | &ip_hdr(skb)->daddr, &gw); |
1da177e4 LT |
864 | #endif |
865 | } | |
1d861aa4 DM |
866 | out_put_peer: |
867 | inet_putpeer(peer); | |
1da177e4 LT |
868 | } |
869 | ||
870 | static int ip_error(struct sk_buff *skb) | |
871 | { | |
251da413 | 872 | struct in_device *in_dev = __in_dev_get_rcu(skb->dev); |
511c3f92 | 873 | struct rtable *rt = skb_rtable(skb); |
92d86829 | 874 | struct inet_peer *peer; |
1da177e4 | 875 | unsigned long now; |
251da413 | 876 | struct net *net; |
92d86829 | 877 | bool send; |
1da177e4 LT |
878 | int code; |
879 | ||
251da413 DM |
880 | net = dev_net(rt->dst.dev); |
881 | if (!IN_DEV_FORWARD(in_dev)) { | |
882 | switch (rt->dst.error) { | |
883 | case EHOSTUNREACH: | |
884 | IP_INC_STATS_BH(net, IPSTATS_MIB_INADDRERRORS); | |
885 | break; | |
886 | ||
887 | case ENETUNREACH: | |
888 | IP_INC_STATS_BH(net, IPSTATS_MIB_INNOROUTES); | |
889 | break; | |
890 | } | |
891 | goto out; | |
892 | } | |
893 | ||
d8d1f30b | 894 | switch (rt->dst.error) { |
4500ebf8 JP |
895 | case EINVAL: |
896 | default: | |
897 | goto out; | |
898 | case EHOSTUNREACH: | |
899 | code = ICMP_HOST_UNREACH; | |
900 | break; | |
901 | case ENETUNREACH: | |
902 | code = ICMP_NET_UNREACH; | |
251da413 | 903 | IP_INC_STATS_BH(net, IPSTATS_MIB_INNOROUTES); |
4500ebf8 JP |
904 | break; |
905 | case EACCES: | |
906 | code = ICMP_PKT_FILTERED; | |
907 | break; | |
1da177e4 LT |
908 | } |
909 | ||
1d861aa4 | 910 | peer = inet_getpeer_v4(net->ipv4.peers, ip_hdr(skb)->saddr, 1); |
92d86829 DM |
911 | |
912 | send = true; | |
913 | if (peer) { | |
914 | now = jiffies; | |
915 | peer->rate_tokens += now - peer->rate_last; | |
916 | if (peer->rate_tokens > ip_rt_error_burst) | |
917 | peer->rate_tokens = ip_rt_error_burst; | |
918 | peer->rate_last = now; | |
919 | if (peer->rate_tokens >= ip_rt_error_cost) | |
920 | peer->rate_tokens -= ip_rt_error_cost; | |
921 | else | |
922 | send = false; | |
1d861aa4 | 923 | inet_putpeer(peer); |
1da177e4 | 924 | } |
92d86829 DM |
925 | if (send) |
926 | icmp_send(skb, ICMP_DEST_UNREACH, code, 0); | |
1da177e4 LT |
927 | |
928 | out: kfree_skb(skb); | |
929 | return 0; | |
e905a9ed | 930 | } |
1da177e4 | 931 | |
d851c12b | 932 | static void __ip_rt_update_pmtu(struct rtable *rt, struct flowi4 *fl4, u32 mtu) |
1da177e4 | 933 | { |
d851c12b | 934 | struct dst_entry *dst = &rt->dst; |
4895c771 | 935 | struct fib_result res; |
2c8cec5c | 936 | |
fa1e492a SK |
937 | if (dst_metric_locked(dst, RTAX_MTU)) |
938 | return; | |
939 | ||
7f92d334 SK |
940 | if (dst->dev->mtu < mtu) |
941 | return; | |
942 | ||
5943634f DM |
943 | if (mtu < ip_rt_min_pmtu) |
944 | mtu = ip_rt_min_pmtu; | |
2c8cec5c | 945 | |
f016229e TT |
946 | if (rt->rt_pmtu == mtu && |
947 | time_before(jiffies, dst->expires - ip_rt_mtu_expires / 2)) | |
948 | return; | |
949 | ||
c5ae7d41 | 950 | rcu_read_lock(); |
d851c12b | 951 | if (fib_lookup(dev_net(dst->dev), fl4, &res) == 0) { |
4895c771 | 952 | struct fib_nh *nh = &FIB_RES_NH(res); |
4895c771 | 953 | |
aee06da6 JA |
954 | update_or_create_fnhe(nh, fl4->daddr, 0, mtu, |
955 | jiffies + ip_rt_mtu_expires); | |
4895c771 | 956 | } |
c5ae7d41 | 957 | rcu_read_unlock(); |
1da177e4 LT |
958 | } |
959 | ||
4895c771 DM |
960 | static void ip_rt_update_pmtu(struct dst_entry *dst, struct sock *sk, |
961 | struct sk_buff *skb, u32 mtu) | |
962 | { | |
963 | struct rtable *rt = (struct rtable *) dst; | |
964 | struct flowi4 fl4; | |
965 | ||
966 | ip_rt_build_flow_key(&fl4, sk, skb); | |
d851c12b | 967 | __ip_rt_update_pmtu(rt, &fl4, mtu); |
4895c771 DM |
968 | } |
969 | ||
36393395 DM |
970 | void ipv4_update_pmtu(struct sk_buff *skb, struct net *net, u32 mtu, |
971 | int oif, u32 mark, u8 protocol, int flow_flags) | |
972 | { | |
4895c771 | 973 | const struct iphdr *iph = (const struct iphdr *) skb->data; |
36393395 DM |
974 | struct flowi4 fl4; |
975 | struct rtable *rt; | |
976 | ||
1b3c61dc LC |
977 | if (!mark) |
978 | mark = IP4_REPLY_MARK(net, skb->mark); | |
979 | ||
4895c771 DM |
980 | __build_flow_key(&fl4, NULL, iph, oif, |
981 | RT_TOS(iph->tos), protocol, mark, flow_flags); | |
36393395 DM |
982 | rt = __ip_route_output_key(net, &fl4); |
983 | if (!IS_ERR(rt)) { | |
4895c771 | 984 | __ip_rt_update_pmtu(rt, &fl4, mtu); |
36393395 DM |
985 | ip_rt_put(rt); |
986 | } | |
987 | } | |
988 | EXPORT_SYMBOL_GPL(ipv4_update_pmtu); | |
989 | ||
9cb3a50c | 990 | static void __ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu) |
36393395 | 991 | { |
4895c771 DM |
992 | const struct iphdr *iph = (const struct iphdr *) skb->data; |
993 | struct flowi4 fl4; | |
994 | struct rtable *rt; | |
36393395 | 995 | |
4895c771 | 996 | __build_flow_key(&fl4, sk, iph, 0, 0, 0, 0, 0); |
1b3c61dc LC |
997 | |
998 | if (!fl4.flowi4_mark) | |
999 | fl4.flowi4_mark = IP4_REPLY_MARK(sock_net(sk), skb->mark); | |
1000 | ||
4895c771 DM |
1001 | rt = __ip_route_output_key(sock_net(sk), &fl4); |
1002 | if (!IS_ERR(rt)) { | |
1003 | __ip_rt_update_pmtu(rt, &fl4, mtu); | |
1004 | ip_rt_put(rt); | |
1005 | } | |
36393395 | 1006 | } |
9cb3a50c SK |
1007 | |
1008 | void ipv4_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, u32 mtu) | |
1009 | { | |
1010 | const struct iphdr *iph = (const struct iphdr *) skb->data; | |
1011 | struct flowi4 fl4; | |
1012 | struct rtable *rt; | |
7f502361 | 1013 | struct dst_entry *odst = NULL; |
b44108db | 1014 | bool new = false; |
9cb3a50c SK |
1015 | |
1016 | bh_lock_sock(sk); | |
482fc609 HFS |
1017 | |
1018 | if (!ip_sk_accept_pmtu(sk)) | |
1019 | goto out; | |
1020 | ||
7f502361 | 1021 | odst = sk_dst_get(sk); |
9cb3a50c | 1022 | |
7f502361 | 1023 | if (sock_owned_by_user(sk) || !odst) { |
9cb3a50c SK |
1024 | __ipv4_sk_update_pmtu(skb, sk, mtu); |
1025 | goto out; | |
1026 | } | |
1027 | ||
1028 | __build_flow_key(&fl4, sk, iph, 0, 0, 0, 0, 0); | |
1029 | ||
7f502361 ED |
1030 | rt = (struct rtable *)odst; |
1031 | if (odst->obsolete && odst->ops->check(odst, 0) == NULL) { | |
9cb3a50c SK |
1032 | rt = ip_route_output_flow(sock_net(sk), &fl4, sk); |
1033 | if (IS_ERR(rt)) | |
1034 | goto out; | |
b44108db SK |
1035 | |
1036 | new = true; | |
9cb3a50c SK |
1037 | } |
1038 | ||
1039 | __ip_rt_update_pmtu((struct rtable *) rt->dst.path, &fl4, mtu); | |
1040 | ||
7f502361 | 1041 | if (!dst_check(&rt->dst, 0)) { |
b44108db SK |
1042 | if (new) |
1043 | dst_release(&rt->dst); | |
1044 | ||
9cb3a50c SK |
1045 | rt = ip_route_output_flow(sock_net(sk), &fl4, sk); |
1046 | if (IS_ERR(rt)) | |
1047 | goto out; | |
1048 | ||
b44108db | 1049 | new = true; |
9cb3a50c SK |
1050 | } |
1051 | ||
b44108db | 1052 | if (new) |
7f502361 | 1053 | sk_dst_set(sk, &rt->dst); |
9cb3a50c SK |
1054 | |
1055 | out: | |
1056 | bh_unlock_sock(sk); | |
7f502361 | 1057 | dst_release(odst); |
9cb3a50c | 1058 | } |
36393395 | 1059 | EXPORT_SYMBOL_GPL(ipv4_sk_update_pmtu); |
f39925db | 1060 | |
b42597e2 DM |
1061 | void ipv4_redirect(struct sk_buff *skb, struct net *net, |
1062 | int oif, u32 mark, u8 protocol, int flow_flags) | |
1063 | { | |
4895c771 | 1064 | const struct iphdr *iph = (const struct iphdr *) skb->data; |
b42597e2 DM |
1065 | struct flowi4 fl4; |
1066 | struct rtable *rt; | |
1067 | ||
4895c771 DM |
1068 | __build_flow_key(&fl4, NULL, iph, oif, |
1069 | RT_TOS(iph->tos), protocol, mark, flow_flags); | |
b42597e2 DM |
1070 | rt = __ip_route_output_key(net, &fl4); |
1071 | if (!IS_ERR(rt)) { | |
ceb33206 | 1072 | __ip_do_redirect(rt, skb, &fl4, false); |
b42597e2 DM |
1073 | ip_rt_put(rt); |
1074 | } | |
1075 | } | |
1076 | EXPORT_SYMBOL_GPL(ipv4_redirect); | |
1077 | ||
1078 | void ipv4_sk_redirect(struct sk_buff *skb, struct sock *sk) | |
1079 | { | |
4895c771 DM |
1080 | const struct iphdr *iph = (const struct iphdr *) skb->data; |
1081 | struct flowi4 fl4; | |
1082 | struct rtable *rt; | |
b42597e2 | 1083 | |
4895c771 DM |
1084 | __build_flow_key(&fl4, sk, iph, 0, 0, 0, 0, 0); |
1085 | rt = __ip_route_output_key(sock_net(sk), &fl4); | |
1086 | if (!IS_ERR(rt)) { | |
ceb33206 | 1087 | __ip_do_redirect(rt, skb, &fl4, false); |
4895c771 DM |
1088 | ip_rt_put(rt); |
1089 | } | |
b42597e2 DM |
1090 | } |
1091 | EXPORT_SYMBOL_GPL(ipv4_sk_redirect); | |
1092 | ||
efbc368d DM |
1093 | static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie) |
1094 | { | |
1095 | struct rtable *rt = (struct rtable *) dst; | |
1096 | ||
ceb33206 DM |
1097 | /* All IPV4 dsts are created with ->obsolete set to the value |
1098 | * DST_OBSOLETE_FORCE_CHK which forces validation calls down | |
1099 | * into this function always. | |
1100 | * | |
387aa65a TT |
1101 | * When a PMTU/redirect information update invalidates a route, |
1102 | * this is indicated by setting obsolete to DST_OBSOLETE_KILL or | |
1103 | * DST_OBSOLETE_DEAD by dst_free(). | |
ceb33206 | 1104 | */ |
387aa65a | 1105 | if (dst->obsolete != DST_OBSOLETE_FORCE_CHK || rt_is_expired(rt)) |
efbc368d | 1106 | return NULL; |
d11a4dc1 | 1107 | return dst; |
1da177e4 LT |
1108 | } |
1109 | ||
1da177e4 LT |
1110 | static void ipv4_link_failure(struct sk_buff *skb) |
1111 | { | |
1112 | struct rtable *rt; | |
1113 | ||
1114 | icmp_send(skb, ICMP_DEST_UNREACH, ICMP_HOST_UNREACH, 0); | |
1115 | ||
511c3f92 | 1116 | rt = skb_rtable(skb); |
5943634f DM |
1117 | if (rt) |
1118 | dst_set_expires(&rt->dst, 0); | |
1da177e4 LT |
1119 | } |
1120 | ||
aad88724 | 1121 | static int ip_rt_bug(struct sock *sk, struct sk_buff *skb) |
1da177e4 | 1122 | { |
91df42be JP |
1123 | pr_debug("%s: %pI4 -> %pI4, %s\n", |
1124 | __func__, &ip_hdr(skb)->saddr, &ip_hdr(skb)->daddr, | |
1125 | skb->dev ? skb->dev->name : "?"); | |
1da177e4 | 1126 | kfree_skb(skb); |
c378a9c0 | 1127 | WARN_ON(1); |
1da177e4 LT |
1128 | return 0; |
1129 | } | |
1130 | ||
1131 | /* | |
1132 | We do not cache source address of outgoing interface, | |
1133 | because it is used only by IP RR, TS and SRR options, | |
1134 | so that it out of fast path. | |
1135 | ||
1136 | BTW remember: "addr" is allowed to be not aligned | |
1137 | in IP options! | |
1138 | */ | |
1139 | ||
8e36360a | 1140 | void ip_rt_get_source(u8 *addr, struct sk_buff *skb, struct rtable *rt) |
1da177e4 | 1141 | { |
a61ced5d | 1142 | __be32 src; |
1da177e4 | 1143 | |
c7537967 | 1144 | if (rt_is_output_route(rt)) |
c5be24ff | 1145 | src = ip_hdr(skb)->saddr; |
ebc0ffae | 1146 | else { |
8e36360a DM |
1147 | struct fib_result res; |
1148 | struct flowi4 fl4; | |
1149 | struct iphdr *iph; | |
1150 | ||
1151 | iph = ip_hdr(skb); | |
1152 | ||
1153 | memset(&fl4, 0, sizeof(fl4)); | |
1154 | fl4.daddr = iph->daddr; | |
1155 | fl4.saddr = iph->saddr; | |
b0fe4a31 | 1156 | fl4.flowi4_tos = RT_TOS(iph->tos); |
8e36360a DM |
1157 | fl4.flowi4_oif = rt->dst.dev->ifindex; |
1158 | fl4.flowi4_iif = skb->dev->ifindex; | |
1159 | fl4.flowi4_mark = skb->mark; | |
5e2b61f7 | 1160 | |
ebc0ffae | 1161 | rcu_read_lock(); |
68a5e3dd | 1162 | if (fib_lookup(dev_net(rt->dst.dev), &fl4, &res) == 0) |
436c3b66 | 1163 | src = FIB_RES_PREFSRC(dev_net(rt->dst.dev), res); |
ebc0ffae | 1164 | else |
f8126f1d DM |
1165 | src = inet_select_addr(rt->dst.dev, |
1166 | rt_nexthop(rt, iph->daddr), | |
1167 | RT_SCOPE_UNIVERSE); | |
ebc0ffae ED |
1168 | rcu_read_unlock(); |
1169 | } | |
1da177e4 LT |
1170 | memcpy(addr, &src, 4); |
1171 | } | |
1172 | ||
c7066f70 | 1173 | #ifdef CONFIG_IP_ROUTE_CLASSID |
1da177e4 LT |
1174 | static void set_class_tag(struct rtable *rt, u32 tag) |
1175 | { | |
d8d1f30b CG |
1176 | if (!(rt->dst.tclassid & 0xFFFF)) |
1177 | rt->dst.tclassid |= tag & 0xFFFF; | |
1178 | if (!(rt->dst.tclassid & 0xFFFF0000)) | |
1179 | rt->dst.tclassid |= tag & 0xFFFF0000; | |
1da177e4 LT |
1180 | } |
1181 | #endif | |
1182 | ||
0dbaee3b DM |
1183 | static unsigned int ipv4_default_advmss(const struct dst_entry *dst) |
1184 | { | |
1185 | unsigned int advmss = dst_metric_raw(dst, RTAX_ADVMSS); | |
1186 | ||
1187 | if (advmss == 0) { | |
1188 | advmss = max_t(unsigned int, dst->dev->mtu - 40, | |
1189 | ip_rt_min_advmss); | |
1190 | if (advmss > 65535 - 40) | |
1191 | advmss = 65535 - 40; | |
1192 | } | |
1193 | return advmss; | |
1194 | } | |
1195 | ||
ebb762f2 | 1196 | static unsigned int ipv4_mtu(const struct dst_entry *dst) |
d33e4553 | 1197 | { |
261663b0 | 1198 | const struct rtable *rt = (const struct rtable *) dst; |
5943634f DM |
1199 | unsigned int mtu = rt->rt_pmtu; |
1200 | ||
98d75c37 | 1201 | if (!mtu || time_after_eq(jiffies, rt->dst.expires)) |
5943634f | 1202 | mtu = dst_metric_raw(dst, RTAX_MTU); |
618f9bc7 | 1203 | |
38d523e2 | 1204 | if (mtu) |
618f9bc7 SK |
1205 | return mtu; |
1206 | ||
1207 | mtu = dst->dev->mtu; | |
d33e4553 DM |
1208 | |
1209 | if (unlikely(dst_metric_locked(dst, RTAX_MTU))) { | |
155e8336 | 1210 | if (rt->rt_uses_gateway && mtu > 576) |
d33e4553 DM |
1211 | mtu = 576; |
1212 | } | |
1213 | ||
734d2725 | 1214 | return min_t(unsigned int, mtu, IP_MAX_MTU); |
d33e4553 DM |
1215 | } |
1216 | ||
f2bb4bed | 1217 | static struct fib_nh_exception *find_exception(struct fib_nh *nh, __be32 daddr) |
4895c771 DM |
1218 | { |
1219 | struct fnhe_hash_bucket *hash = nh->nh_exceptions; | |
1220 | struct fib_nh_exception *fnhe; | |
1221 | u32 hval; | |
1222 | ||
f2bb4bed DM |
1223 | if (!hash) |
1224 | return NULL; | |
1225 | ||
d3a25c98 | 1226 | hval = fnhe_hashfun(daddr); |
4895c771 DM |
1227 | |
1228 | for (fnhe = rcu_dereference(hash[hval].chain); fnhe; | |
1229 | fnhe = rcu_dereference(fnhe->fnhe_next)) { | |
f2bb4bed DM |
1230 | if (fnhe->fnhe_daddr == daddr) |
1231 | return fnhe; | |
1232 | } | |
1233 | return NULL; | |
1234 | } | |
aee06da6 | 1235 | |
caacf05e | 1236 | static bool rt_bind_exception(struct rtable *rt, struct fib_nh_exception *fnhe, |
f2bb4bed DM |
1237 | __be32 daddr) |
1238 | { | |
caacf05e DM |
1239 | bool ret = false; |
1240 | ||
c5038a83 | 1241 | spin_lock_bh(&fnhe_lock); |
f2bb4bed | 1242 | |
c5038a83 | 1243 | if (daddr == fnhe->fnhe_daddr) { |
2ffae99d TT |
1244 | struct rtable __rcu **porig; |
1245 | struct rtable *orig; | |
5aad1de5 | 1246 | int genid = fnhe_genid(dev_net(rt->dst.dev)); |
2ffae99d TT |
1247 | |
1248 | if (rt_is_input_route(rt)) | |
1249 | porig = &fnhe->fnhe_rth_input; | |
1250 | else | |
1251 | porig = &fnhe->fnhe_rth_output; | |
1252 | orig = rcu_dereference(*porig); | |
5aad1de5 TT |
1253 | |
1254 | if (fnhe->fnhe_genid != genid) { | |
1255 | fnhe->fnhe_genid = genid; | |
13d82bf5 SK |
1256 | fnhe->fnhe_gw = 0; |
1257 | fnhe->fnhe_pmtu = 0; | |
1258 | fnhe->fnhe_expires = 0; | |
2ffae99d TT |
1259 | fnhe_flush_routes(fnhe); |
1260 | orig = NULL; | |
13d82bf5 | 1261 | } |
387aa65a TT |
1262 | fill_route_from_fnhe(rt, fnhe); |
1263 | if (!rt->rt_gateway) | |
155e8336 | 1264 | rt->rt_gateway = daddr; |
f2bb4bed | 1265 | |
2ffae99d TT |
1266 | if (!(rt->dst.flags & DST_NOCACHE)) { |
1267 | rcu_assign_pointer(*porig, rt); | |
1268 | if (orig) | |
1269 | rt_free(orig); | |
1270 | ret = true; | |
1271 | } | |
c5038a83 DM |
1272 | |
1273 | fnhe->fnhe_stamp = jiffies; | |
c5038a83 DM |
1274 | } |
1275 | spin_unlock_bh(&fnhe_lock); | |
caacf05e DM |
1276 | |
1277 | return ret; | |
54764bb6 ED |
1278 | } |
1279 | ||
caacf05e | 1280 | static bool rt_cache_route(struct fib_nh *nh, struct rtable *rt) |
f2bb4bed | 1281 | { |
d26b3a7c | 1282 | struct rtable *orig, *prev, **p; |
caacf05e | 1283 | bool ret = true; |
f2bb4bed | 1284 | |
d26b3a7c | 1285 | if (rt_is_input_route(rt)) { |
54764bb6 | 1286 | p = (struct rtable **)&nh->nh_rth_input; |
d26b3a7c | 1287 | } else { |
d26b3a7c ED |
1288 | p = (struct rtable **)__this_cpu_ptr(nh->nh_pcpu_rth_output); |
1289 | } | |
f2bb4bed DM |
1290 | orig = *p; |
1291 | ||
1292 | prev = cmpxchg(p, orig, rt); | |
1293 | if (prev == orig) { | |
f2bb4bed | 1294 | if (orig) |
54764bb6 | 1295 | rt_free(orig); |
155e8336 | 1296 | } else |
caacf05e | 1297 | ret = false; |
caacf05e DM |
1298 | |
1299 | return ret; | |
1300 | } | |
1301 | ||
1302 | static DEFINE_SPINLOCK(rt_uncached_lock); | |
1303 | static LIST_HEAD(rt_uncached_list); | |
1304 | ||
1305 | static void rt_add_uncached_list(struct rtable *rt) | |
1306 | { | |
1307 | spin_lock_bh(&rt_uncached_lock); | |
1308 | list_add_tail(&rt->rt_uncached, &rt_uncached_list); | |
1309 | spin_unlock_bh(&rt_uncached_lock); | |
1310 | } | |
1311 | ||
1312 | static void ipv4_dst_destroy(struct dst_entry *dst) | |
1313 | { | |
1314 | struct rtable *rt = (struct rtable *) dst; | |
1315 | ||
78df76a0 | 1316 | if (!list_empty(&rt->rt_uncached)) { |
caacf05e DM |
1317 | spin_lock_bh(&rt_uncached_lock); |
1318 | list_del(&rt->rt_uncached); | |
1319 | spin_unlock_bh(&rt_uncached_lock); | |
1320 | } | |
1321 | } | |
1322 | ||
1323 | void rt_flush_dev(struct net_device *dev) | |
1324 | { | |
1325 | if (!list_empty(&rt_uncached_list)) { | |
1326 | struct net *net = dev_net(dev); | |
1327 | struct rtable *rt; | |
1328 | ||
1329 | spin_lock_bh(&rt_uncached_lock); | |
1330 | list_for_each_entry(rt, &rt_uncached_list, rt_uncached) { | |
1331 | if (rt->dst.dev != dev) | |
1332 | continue; | |
1333 | rt->dst.dev = net->loopback_dev; | |
1334 | dev_hold(rt->dst.dev); | |
1335 | dev_put(dev); | |
1336 | } | |
1337 | spin_unlock_bh(&rt_uncached_lock); | |
4895c771 DM |
1338 | } |
1339 | } | |
1340 | ||
4331debc | 1341 | static bool rt_cache_valid(const struct rtable *rt) |
d2d68ba9 | 1342 | { |
4331debc ED |
1343 | return rt && |
1344 | rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK && | |
1345 | !rt_is_expired(rt); | |
d2d68ba9 DM |
1346 | } |
1347 | ||
f2bb4bed | 1348 | static void rt_set_nexthop(struct rtable *rt, __be32 daddr, |
5e2b61f7 | 1349 | const struct fib_result *res, |
f2bb4bed | 1350 | struct fib_nh_exception *fnhe, |
982721f3 | 1351 | struct fib_info *fi, u16 type, u32 itag) |
1da177e4 | 1352 | { |
caacf05e DM |
1353 | bool cached = false; |
1354 | ||
1da177e4 | 1355 | if (fi) { |
4895c771 DM |
1356 | struct fib_nh *nh = &FIB_RES_NH(*res); |
1357 | ||
155e8336 | 1358 | if (nh->nh_gw && nh->nh_scope == RT_SCOPE_LINK) { |
4895c771 | 1359 | rt->rt_gateway = nh->nh_gw; |
155e8336 JA |
1360 | rt->rt_uses_gateway = 1; |
1361 | } | |
2860583f | 1362 | dst_init_metrics(&rt->dst, fi->fib_metrics, true); |
c7066f70 | 1363 | #ifdef CONFIG_IP_ROUTE_CLASSID |
f2bb4bed | 1364 | rt->dst.tclassid = nh->nh_tclassid; |
1da177e4 | 1365 | #endif |
c5038a83 | 1366 | if (unlikely(fnhe)) |
caacf05e | 1367 | cached = rt_bind_exception(rt, fnhe, daddr); |
c5038a83 | 1368 | else if (!(rt->dst.flags & DST_NOCACHE)) |
caacf05e | 1369 | cached = rt_cache_route(nh, rt); |
155e8336 JA |
1370 | if (unlikely(!cached)) { |
1371 | /* Routes we intend to cache in nexthop exception or | |
1372 | * FIB nexthop have the DST_NOCACHE bit clear. | |
1373 | * However, if we are unsuccessful at storing this | |
1374 | * route into the cache we really need to set it. | |
1375 | */ | |
1376 | rt->dst.flags |= DST_NOCACHE; | |
1377 | if (!rt->rt_gateway) | |
1378 | rt->rt_gateway = daddr; | |
1379 | rt_add_uncached_list(rt); | |
1380 | } | |
1381 | } else | |
caacf05e | 1382 | rt_add_uncached_list(rt); |
defb3519 | 1383 | |
c7066f70 | 1384 | #ifdef CONFIG_IP_ROUTE_CLASSID |
1da177e4 | 1385 | #ifdef CONFIG_IP_MULTIPLE_TABLES |
85b91b03 | 1386 | set_class_tag(rt, res->tclassid); |
1da177e4 LT |
1387 | #endif |
1388 | set_class_tag(rt, itag); | |
1389 | #endif | |
1da177e4 LT |
1390 | } |
1391 | ||
5c1e6aa3 | 1392 | static struct rtable *rt_dst_alloc(struct net_device *dev, |
f2bb4bed | 1393 | bool nopolicy, bool noxfrm, bool will_cache) |
0c4dcd58 | 1394 | { |
f5b0a874 | 1395 | return dst_alloc(&ipv4_dst_ops, dev, 1, DST_OBSOLETE_FORCE_CHK, |
c6cffba4 | 1396 | (will_cache ? 0 : (DST_HOST | DST_NOCACHE)) | |
5c1e6aa3 DM |
1397 | (nopolicy ? DST_NOPOLICY : 0) | |
1398 | (noxfrm ? DST_NOXFRM : 0)); | |
0c4dcd58 DM |
1399 | } |
1400 | ||
96d36220 | 1401 | /* called in rcu_read_lock() section */ |
9e12bb22 | 1402 | static int ip_route_input_mc(struct sk_buff *skb, __be32 daddr, __be32 saddr, |
1da177e4 LT |
1403 | u8 tos, struct net_device *dev, int our) |
1404 | { | |
1da177e4 | 1405 | struct rtable *rth; |
96d36220 | 1406 | struct in_device *in_dev = __in_dev_get_rcu(dev); |
1da177e4 | 1407 | u32 itag = 0; |
b5f7e755 | 1408 | int err; |
1da177e4 LT |
1409 | |
1410 | /* Primary sanity checks. */ | |
1411 | ||
1412 | if (in_dev == NULL) | |
1413 | return -EINVAL; | |
1414 | ||
1e637c74 | 1415 | if (ipv4_is_multicast(saddr) || ipv4_is_lbcast(saddr) || |
d0daebc3 | 1416 | skb->protocol != htons(ETH_P_IP)) |
1da177e4 LT |
1417 | goto e_inval; |
1418 | ||
d0daebc3 TG |
1419 | if (likely(!IN_DEV_ROUTE_LOCALNET(in_dev))) |
1420 | if (ipv4_is_loopback(saddr)) | |
1421 | goto e_inval; | |
1422 | ||
f97c1e0c JP |
1423 | if (ipv4_is_zeronet(saddr)) { |
1424 | if (!ipv4_is_local_multicast(daddr)) | |
1da177e4 | 1425 | goto e_inval; |
b5f7e755 | 1426 | } else { |
9e56e380 DM |
1427 | err = fib_validate_source(skb, saddr, 0, tos, 0, dev, |
1428 | in_dev, &itag); | |
b5f7e755 ED |
1429 | if (err < 0) |
1430 | goto e_err; | |
1431 | } | |
4e7b2f14 | 1432 | rth = rt_dst_alloc(dev_net(dev)->loopback_dev, |
f2bb4bed | 1433 | IN_DEV_CONF_GET(in_dev, NOPOLICY), false, false); |
1da177e4 LT |
1434 | if (!rth) |
1435 | goto e_nobufs; | |
1436 | ||
cf911662 DM |
1437 | #ifdef CONFIG_IP_ROUTE_CLASSID |
1438 | rth->dst.tclassid = itag; | |
1439 | #endif | |
d8d1f30b | 1440 | rth->dst.output = ip_rt_bug; |
1da177e4 | 1441 | |
ca4c3fc2 | 1442 | rth->rt_genid = rt_genid_ipv4(dev_net(dev)); |
cf911662 DM |
1443 | rth->rt_flags = RTCF_MULTICAST; |
1444 | rth->rt_type = RTN_MULTICAST; | |
9917e1e8 | 1445 | rth->rt_is_input= 1; |
13378cad | 1446 | rth->rt_iif = 0; |
5943634f | 1447 | rth->rt_pmtu = 0; |
f8126f1d | 1448 | rth->rt_gateway = 0; |
155e8336 | 1449 | rth->rt_uses_gateway = 0; |
caacf05e | 1450 | INIT_LIST_HEAD(&rth->rt_uncached); |
1da177e4 | 1451 | if (our) { |
d8d1f30b | 1452 | rth->dst.input= ip_local_deliver; |
1da177e4 LT |
1453 | rth->rt_flags |= RTCF_LOCAL; |
1454 | } | |
1455 | ||
1456 | #ifdef CONFIG_IP_MROUTE | |
f97c1e0c | 1457 | if (!ipv4_is_local_multicast(daddr) && IN_DEV_MFORWARD(in_dev)) |
d8d1f30b | 1458 | rth->dst.input = ip_mr_input; |
1da177e4 LT |
1459 | #endif |
1460 | RT_CACHE_STAT_INC(in_slow_mc); | |
1461 | ||
89aef892 DM |
1462 | skb_dst_set(skb, &rth->dst); |
1463 | return 0; | |
1da177e4 LT |
1464 | |
1465 | e_nobufs: | |
1da177e4 | 1466 | return -ENOBUFS; |
1da177e4 | 1467 | e_inval: |
96d36220 | 1468 | return -EINVAL; |
b5f7e755 | 1469 | e_err: |
b5f7e755 | 1470 | return err; |
1da177e4 LT |
1471 | } |
1472 | ||
1473 | ||
1474 | static void ip_handle_martian_source(struct net_device *dev, | |
1475 | struct in_device *in_dev, | |
1476 | struct sk_buff *skb, | |
9e12bb22 AV |
1477 | __be32 daddr, |
1478 | __be32 saddr) | |
1da177e4 LT |
1479 | { |
1480 | RT_CACHE_STAT_INC(in_martian_src); | |
1481 | #ifdef CONFIG_IP_ROUTE_VERBOSE | |
1482 | if (IN_DEV_LOG_MARTIANS(in_dev) && net_ratelimit()) { | |
1483 | /* | |
1484 | * RFC1812 recommendation, if source is martian, | |
1485 | * the only hint is MAC header. | |
1486 | */ | |
058bd4d2 | 1487 | pr_warn("martian source %pI4 from %pI4, on dev %s\n", |
673d57e7 | 1488 | &daddr, &saddr, dev->name); |
98e399f8 | 1489 | if (dev->hard_header_len && skb_mac_header_was_set(skb)) { |
058bd4d2 JP |
1490 | print_hex_dump(KERN_WARNING, "ll header: ", |
1491 | DUMP_PREFIX_OFFSET, 16, 1, | |
1492 | skb_mac_header(skb), | |
1493 | dev->hard_header_len, true); | |
1da177e4 LT |
1494 | } |
1495 | } | |
1496 | #endif | |
1497 | } | |
1498 | ||
47360228 | 1499 | /* called in rcu_read_lock() section */ |
5969f71d | 1500 | static int __mkroute_input(struct sk_buff *skb, |
982721f3 | 1501 | const struct fib_result *res, |
5969f71d | 1502 | struct in_device *in_dev, |
c6cffba4 | 1503 | __be32 daddr, __be32 saddr, u32 tos) |
1da177e4 | 1504 | { |
2ffae99d | 1505 | struct fib_nh_exception *fnhe; |
1da177e4 LT |
1506 | struct rtable *rth; |
1507 | int err; | |
1508 | struct in_device *out_dev; | |
47360228 | 1509 | unsigned int flags = 0; |
d2d68ba9 | 1510 | bool do_cache; |
fbdc0ad0 | 1511 | u32 itag = 0; |
1da177e4 LT |
1512 | |
1513 | /* get a working reference to the output device */ | |
47360228 | 1514 | out_dev = __in_dev_get_rcu(FIB_RES_DEV(*res)); |
1da177e4 | 1515 | if (out_dev == NULL) { |
e87cc472 | 1516 | net_crit_ratelimited("Bug in ip_route_input_slow(). Please report.\n"); |
1da177e4 LT |
1517 | return -EINVAL; |
1518 | } | |
1519 | ||
5c04c819 | 1520 | err = fib_validate_source(skb, saddr, daddr, tos, FIB_RES_OIF(*res), |
9e56e380 | 1521 | in_dev->dev, in_dev, &itag); |
1da177e4 | 1522 | if (err < 0) { |
e905a9ed | 1523 | ip_handle_martian_source(in_dev->dev, in_dev, skb, daddr, |
1da177e4 | 1524 | saddr); |
e905a9ed | 1525 | |
1da177e4 LT |
1526 | goto cleanup; |
1527 | } | |
1528 | ||
e81da0e1 JA |
1529 | do_cache = res->fi && !itag; |
1530 | if (out_dev == in_dev && err && IN_DEV_TX_REDIRECTS(out_dev) && | |
1da177e4 | 1531 | (IN_DEV_SHARED_MEDIA(out_dev) || |
e81da0e1 | 1532 | inet_addr_onlink(out_dev, saddr, FIB_RES_GW(*res)))) { |
1da177e4 | 1533 | flags |= RTCF_DOREDIRECT; |
e81da0e1 JA |
1534 | do_cache = false; |
1535 | } | |
1da177e4 LT |
1536 | |
1537 | if (skb->protocol != htons(ETH_P_IP)) { | |
1538 | /* Not IP (i.e. ARP). Do not create route, if it is | |
1539 | * invalid for proxy arp. DNAT routes are always valid. | |
65324144 JDB |
1540 | * |
1541 | * Proxy arp feature have been extended to allow, ARP | |
1542 | * replies back to the same interface, to support | |
1543 | * Private VLAN switch technologies. See arp.c. | |
1da177e4 | 1544 | */ |
65324144 JDB |
1545 | if (out_dev == in_dev && |
1546 | IN_DEV_PROXY_ARP_PVLAN(in_dev) == 0) { | |
1da177e4 LT |
1547 | err = -EINVAL; |
1548 | goto cleanup; | |
1549 | } | |
1550 | } | |
1551 | ||
2ffae99d | 1552 | fnhe = find_exception(&FIB_RES_NH(*res), daddr); |
e81da0e1 | 1553 | if (do_cache) { |
2ffae99d TT |
1554 | if (fnhe != NULL) |
1555 | rth = rcu_dereference(fnhe->fnhe_rth_input); | |
1556 | else | |
1557 | rth = rcu_dereference(FIB_RES_NH(*res).nh_rth_input); | |
1558 | ||
e81da0e1 JA |
1559 | if (rt_cache_valid(rth)) { |
1560 | skb_dst_set_noref(skb, &rth->dst); | |
1561 | goto out; | |
d2d68ba9 DM |
1562 | } |
1563 | } | |
f2bb4bed | 1564 | |
5c1e6aa3 DM |
1565 | rth = rt_dst_alloc(out_dev->dev, |
1566 | IN_DEV_CONF_GET(in_dev, NOPOLICY), | |
d2d68ba9 | 1567 | IN_DEV_CONF_GET(out_dev, NOXFRM), do_cache); |
1da177e4 LT |
1568 | if (!rth) { |
1569 | err = -ENOBUFS; | |
1570 | goto cleanup; | |
1571 | } | |
1572 | ||
ca4c3fc2 | 1573 | rth->rt_genid = rt_genid_ipv4(dev_net(rth->dst.dev)); |
cf911662 DM |
1574 | rth->rt_flags = flags; |
1575 | rth->rt_type = res->type; | |
9917e1e8 | 1576 | rth->rt_is_input = 1; |
13378cad | 1577 | rth->rt_iif = 0; |
5943634f | 1578 | rth->rt_pmtu = 0; |
f8126f1d | 1579 | rth->rt_gateway = 0; |
155e8336 | 1580 | rth->rt_uses_gateway = 0; |
caacf05e | 1581 | INIT_LIST_HEAD(&rth->rt_uncached); |
a6254864 | 1582 | RT_CACHE_STAT_INC(in_slow_tot); |
1da177e4 | 1583 | |
d8d1f30b CG |
1584 | rth->dst.input = ip_forward; |
1585 | rth->dst.output = ip_output; | |
1da177e4 | 1586 | |
2ffae99d | 1587 | rt_set_nexthop(rth, daddr, res, fnhe, res->fi, res->type, itag); |
c6cffba4 | 1588 | skb_dst_set(skb, &rth->dst); |
d2d68ba9 | 1589 | out: |
1da177e4 LT |
1590 | err = 0; |
1591 | cleanup: | |
1da177e4 | 1592 | return err; |
e905a9ed | 1593 | } |
1da177e4 | 1594 | |
5969f71d SH |
1595 | static int ip_mkroute_input(struct sk_buff *skb, |
1596 | struct fib_result *res, | |
68a5e3dd | 1597 | const struct flowi4 *fl4, |
5969f71d SH |
1598 | struct in_device *in_dev, |
1599 | __be32 daddr, __be32 saddr, u32 tos) | |
1da177e4 | 1600 | { |
1da177e4 | 1601 | #ifdef CONFIG_IP_ROUTE_MULTIPATH |
ff3fccb3 | 1602 | if (res->fi && res->fi->fib_nhs > 1) |
1b7fe593 | 1603 | fib_select_multipath(res); |
1da177e4 LT |
1604 | #endif |
1605 | ||
1606 | /* create a routing cache entry */ | |
c6cffba4 | 1607 | return __mkroute_input(skb, res, in_dev, daddr, saddr, tos); |
1da177e4 LT |
1608 | } |
1609 | ||
1da177e4 LT |
1610 | /* |
1611 | * NOTE. We drop all the packets that has local source | |
1612 | * addresses, because every properly looped back packet | |
1613 | * must have correct destination already attached by output routine. | |
1614 | * | |
1615 | * Such approach solves two big problems: | |
1616 | * 1. Not simplex devices are handled properly. | |
1617 | * 2. IP spoofing attempts are filtered with 100% of guarantee. | |
ebc0ffae | 1618 | * called with rcu_read_lock() |
1da177e4 LT |
1619 | */ |
1620 | ||
9e12bb22 | 1621 | static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr, |
c10237e0 | 1622 | u8 tos, struct net_device *dev) |
1da177e4 LT |
1623 | { |
1624 | struct fib_result res; | |
96d36220 | 1625 | struct in_device *in_dev = __in_dev_get_rcu(dev); |
68a5e3dd | 1626 | struct flowi4 fl4; |
95c96174 | 1627 | unsigned int flags = 0; |
1da177e4 | 1628 | u32 itag = 0; |
95c96174 | 1629 | struct rtable *rth; |
1da177e4 | 1630 | int err = -EINVAL; |
5e73ea1a | 1631 | struct net *net = dev_net(dev); |
d2d68ba9 | 1632 | bool do_cache; |
1da177e4 LT |
1633 | |
1634 | /* IP on this device is disabled. */ | |
1635 | ||
1636 | if (!in_dev) | |
1637 | goto out; | |
1638 | ||
1639 | /* Check for the most weird martians, which can be not detected | |
1640 | by fib_lookup. | |
1641 | */ | |
1642 | ||
d0daebc3 | 1643 | if (ipv4_is_multicast(saddr) || ipv4_is_lbcast(saddr)) |
1da177e4 LT |
1644 | goto martian_source; |
1645 | ||
d2d68ba9 | 1646 | res.fi = NULL; |
27a954bd | 1647 | if (ipv4_is_lbcast(daddr) || (saddr == 0 && daddr == 0)) |
1da177e4 LT |
1648 | goto brd_input; |
1649 | ||
1650 | /* Accept zero addresses only to limited broadcast; | |
1651 | * I even do not know to fix it or not. Waiting for complains :-) | |
1652 | */ | |
f97c1e0c | 1653 | if (ipv4_is_zeronet(saddr)) |
1da177e4 LT |
1654 | goto martian_source; |
1655 | ||
d0daebc3 | 1656 | if (ipv4_is_zeronet(daddr)) |
1da177e4 LT |
1657 | goto martian_destination; |
1658 | ||
9eb43e76 ED |
1659 | /* Following code try to avoid calling IN_DEV_NET_ROUTE_LOCALNET(), |
1660 | * and call it once if daddr or/and saddr are loopback addresses | |
1661 | */ | |
1662 | if (ipv4_is_loopback(daddr)) { | |
1663 | if (!IN_DEV_NET_ROUTE_LOCALNET(in_dev, net)) | |
d0daebc3 | 1664 | goto martian_destination; |
9eb43e76 ED |
1665 | } else if (ipv4_is_loopback(saddr)) { |
1666 | if (!IN_DEV_NET_ROUTE_LOCALNET(in_dev, net)) | |
d0daebc3 TG |
1667 | goto martian_source; |
1668 | } | |
1669 | ||
1da177e4 LT |
1670 | /* |
1671 | * Now we are ready to route packet. | |
1672 | */ | |
68a5e3dd DM |
1673 | fl4.flowi4_oif = 0; |
1674 | fl4.flowi4_iif = dev->ifindex; | |
1675 | fl4.flowi4_mark = skb->mark; | |
1676 | fl4.flowi4_tos = tos; | |
1677 | fl4.flowi4_scope = RT_SCOPE_UNIVERSE; | |
1678 | fl4.daddr = daddr; | |
1679 | fl4.saddr = saddr; | |
1680 | err = fib_lookup(net, &fl4, &res); | |
cd0f0b95 DJ |
1681 | if (err != 0) { |
1682 | if (!IN_DEV_FORWARD(in_dev)) | |
1683 | err = -EHOSTUNREACH; | |
1da177e4 | 1684 | goto no_route; |
cd0f0b95 | 1685 | } |
1da177e4 LT |
1686 | |
1687 | if (res.type == RTN_BROADCAST) | |
1688 | goto brd_input; | |
1689 | ||
1690 | if (res.type == RTN_LOCAL) { | |
5c04c819 | 1691 | err = fib_validate_source(skb, saddr, daddr, tos, |
0d5edc68 | 1692 | 0, dev, in_dev, &itag); |
b5f7e755 ED |
1693 | if (err < 0) |
1694 | goto martian_source_keep_err; | |
1da177e4 LT |
1695 | goto local_input; |
1696 | } | |
1697 | ||
cd0f0b95 DJ |
1698 | if (!IN_DEV_FORWARD(in_dev)) { |
1699 | err = -EHOSTUNREACH; | |
251da413 | 1700 | goto no_route; |
cd0f0b95 | 1701 | } |
1da177e4 LT |
1702 | if (res.type != RTN_UNICAST) |
1703 | goto martian_destination; | |
1704 | ||
68a5e3dd | 1705 | err = ip_mkroute_input(skb, &res, &fl4, in_dev, daddr, saddr, tos); |
1da177e4 LT |
1706 | out: return err; |
1707 | ||
1708 | brd_input: | |
1709 | if (skb->protocol != htons(ETH_P_IP)) | |
1710 | goto e_inval; | |
1711 | ||
41347dcd | 1712 | if (!ipv4_is_zeronet(saddr)) { |
9e56e380 DM |
1713 | err = fib_validate_source(skb, saddr, 0, tos, 0, dev, |
1714 | in_dev, &itag); | |
1da177e4 | 1715 | if (err < 0) |
b5f7e755 | 1716 | goto martian_source_keep_err; |
1da177e4 LT |
1717 | } |
1718 | flags |= RTCF_BROADCAST; | |
1719 | res.type = RTN_BROADCAST; | |
1720 | RT_CACHE_STAT_INC(in_brd); | |
1721 | ||
1722 | local_input: | |
d2d68ba9 DM |
1723 | do_cache = false; |
1724 | if (res.fi) { | |
fe3edf45 | 1725 | if (!itag) { |
54764bb6 | 1726 | rth = rcu_dereference(FIB_RES_NH(res).nh_rth_input); |
d2d68ba9 | 1727 | if (rt_cache_valid(rth)) { |
c6cffba4 DM |
1728 | skb_dst_set_noref(skb, &rth->dst); |
1729 | err = 0; | |
1730 | goto out; | |
d2d68ba9 DM |
1731 | } |
1732 | do_cache = true; | |
1733 | } | |
1734 | } | |
1735 | ||
5c1e6aa3 | 1736 | rth = rt_dst_alloc(net->loopback_dev, |
d2d68ba9 | 1737 | IN_DEV_CONF_GET(in_dev, NOPOLICY), false, do_cache); |
1da177e4 LT |
1738 | if (!rth) |
1739 | goto e_nobufs; | |
1740 | ||
cf911662 | 1741 | rth->dst.input= ip_local_deliver; |
d8d1f30b | 1742 | rth->dst.output= ip_rt_bug; |
cf911662 DM |
1743 | #ifdef CONFIG_IP_ROUTE_CLASSID |
1744 | rth->dst.tclassid = itag; | |
1745 | #endif | |
1da177e4 | 1746 | |
ca4c3fc2 | 1747 | rth->rt_genid = rt_genid_ipv4(net); |
cf911662 DM |
1748 | rth->rt_flags = flags|RTCF_LOCAL; |
1749 | rth->rt_type = res.type; | |
9917e1e8 | 1750 | rth->rt_is_input = 1; |
13378cad | 1751 | rth->rt_iif = 0; |
5943634f | 1752 | rth->rt_pmtu = 0; |
f8126f1d | 1753 | rth->rt_gateway = 0; |
155e8336 | 1754 | rth->rt_uses_gateway = 0; |
caacf05e | 1755 | INIT_LIST_HEAD(&rth->rt_uncached); |
a6254864 | 1756 | RT_CACHE_STAT_INC(in_slow_tot); |
1da177e4 | 1757 | if (res.type == RTN_UNREACHABLE) { |
d8d1f30b CG |
1758 | rth->dst.input= ip_error; |
1759 | rth->dst.error= -err; | |
1da177e4 LT |
1760 | rth->rt_flags &= ~RTCF_LOCAL; |
1761 | } | |
dcdfdf56 AS |
1762 | if (do_cache) { |
1763 | if (unlikely(!rt_cache_route(&FIB_RES_NH(res), rth))) { | |
1764 | rth->dst.flags |= DST_NOCACHE; | |
1765 | rt_add_uncached_list(rth); | |
1766 | } | |
1767 | } | |
89aef892 | 1768 | skb_dst_set(skb, &rth->dst); |
b23dd4fe | 1769 | err = 0; |
ebc0ffae | 1770 | goto out; |
1da177e4 LT |
1771 | |
1772 | no_route: | |
1773 | RT_CACHE_STAT_INC(in_no_route); | |
1da177e4 | 1774 | res.type = RTN_UNREACHABLE; |
7f53878d MC |
1775 | if (err == -ESRCH) |
1776 | err = -ENETUNREACH; | |
1da177e4 LT |
1777 | goto local_input; |
1778 | ||
1779 | /* | |
1780 | * Do not cache martian addresses: they should be logged (RFC1812) | |
1781 | */ | |
1782 | martian_destination: | |
1783 | RT_CACHE_STAT_INC(in_martian_dst); | |
1784 | #ifdef CONFIG_IP_ROUTE_VERBOSE | |
e87cc472 JP |
1785 | if (IN_DEV_LOG_MARTIANS(in_dev)) |
1786 | net_warn_ratelimited("martian destination %pI4 from %pI4, dev %s\n", | |
1787 | &daddr, &saddr, dev->name); | |
1da177e4 | 1788 | #endif |
2c2910a4 | 1789 | |
1da177e4 LT |
1790 | e_inval: |
1791 | err = -EINVAL; | |
ebc0ffae | 1792 | goto out; |
1da177e4 LT |
1793 | |
1794 | e_nobufs: | |
1795 | err = -ENOBUFS; | |
ebc0ffae | 1796 | goto out; |
1da177e4 LT |
1797 | |
1798 | martian_source: | |
b5f7e755 ED |
1799 | err = -EINVAL; |
1800 | martian_source_keep_err: | |
1da177e4 | 1801 | ip_handle_martian_source(dev, in_dev, skb, daddr, saddr); |
ebc0ffae | 1802 | goto out; |
1da177e4 LT |
1803 | } |
1804 | ||
c6cffba4 DM |
1805 | int ip_route_input_noref(struct sk_buff *skb, __be32 daddr, __be32 saddr, |
1806 | u8 tos, struct net_device *dev) | |
1da177e4 | 1807 | { |
96d36220 | 1808 | int res; |
1da177e4 | 1809 | |
96d36220 ED |
1810 | rcu_read_lock(); |
1811 | ||
1da177e4 LT |
1812 | /* Multicast recognition logic is moved from route cache to here. |
1813 | The problem was that too many Ethernet cards have broken/missing | |
1814 | hardware multicast filters :-( As result the host on multicasting | |
1815 | network acquires a lot of useless route cache entries, sort of | |
1816 | SDR messages from all the world. Now we try to get rid of them. | |
1817 | Really, provided software IP multicast filter is organized | |
1818 | reasonably (at least, hashed), it does not result in a slowdown | |
1819 | comparing with route cache reject entries. | |
1820 | Note, that multicast routers are not affected, because | |
1821 | route cache entry is created eventually. | |
1822 | */ | |
f97c1e0c | 1823 | if (ipv4_is_multicast(daddr)) { |
96d36220 | 1824 | struct in_device *in_dev = __in_dev_get_rcu(dev); |
1da177e4 | 1825 | |
96d36220 | 1826 | if (in_dev) { |
dbdd9a52 DM |
1827 | int our = ip_check_mc_rcu(in_dev, daddr, saddr, |
1828 | ip_hdr(skb)->protocol); | |
1da177e4 LT |
1829 | if (our |
1830 | #ifdef CONFIG_IP_MROUTE | |
9d4fb27d JP |
1831 | || |
1832 | (!ipv4_is_local_multicast(daddr) && | |
1833 | IN_DEV_MFORWARD(in_dev)) | |
1da177e4 | 1834 | #endif |
9d4fb27d | 1835 | ) { |
96d36220 ED |
1836 | int res = ip_route_input_mc(skb, daddr, saddr, |
1837 | tos, dev, our); | |
1da177e4 | 1838 | rcu_read_unlock(); |
96d36220 | 1839 | return res; |
1da177e4 LT |
1840 | } |
1841 | } | |
1842 | rcu_read_unlock(); | |
1843 | return -EINVAL; | |
1844 | } | |
c10237e0 | 1845 | res = ip_route_input_slow(skb, daddr, saddr, tos, dev); |
96d36220 ED |
1846 | rcu_read_unlock(); |
1847 | return res; | |
1da177e4 | 1848 | } |
c6cffba4 | 1849 | EXPORT_SYMBOL(ip_route_input_noref); |
1da177e4 | 1850 | |
ebc0ffae | 1851 | /* called with rcu_read_lock() */ |
982721f3 | 1852 | static struct rtable *__mkroute_output(const struct fib_result *res, |
1a00fee4 | 1853 | const struct flowi4 *fl4, int orig_oif, |
f61759e6 | 1854 | struct net_device *dev_out, |
5ada5527 | 1855 | unsigned int flags) |
1da177e4 | 1856 | { |
982721f3 | 1857 | struct fib_info *fi = res->fi; |
f2bb4bed | 1858 | struct fib_nh_exception *fnhe; |
5ada5527 | 1859 | struct in_device *in_dev; |
982721f3 | 1860 | u16 type = res->type; |
5ada5527 | 1861 | struct rtable *rth; |
c92b9655 | 1862 | bool do_cache; |
1da177e4 | 1863 | |
d0daebc3 TG |
1864 | in_dev = __in_dev_get_rcu(dev_out); |
1865 | if (!in_dev) | |
5ada5527 | 1866 | return ERR_PTR(-EINVAL); |
1da177e4 | 1867 | |
d0daebc3 TG |
1868 | if (likely(!IN_DEV_ROUTE_LOCALNET(in_dev))) |
1869 | if (ipv4_is_loopback(fl4->saddr) && !(dev_out->flags & IFF_LOOPBACK)) | |
1870 | return ERR_PTR(-EINVAL); | |
1871 | ||
68a5e3dd | 1872 | if (ipv4_is_lbcast(fl4->daddr)) |
982721f3 | 1873 | type = RTN_BROADCAST; |
68a5e3dd | 1874 | else if (ipv4_is_multicast(fl4->daddr)) |
982721f3 | 1875 | type = RTN_MULTICAST; |
68a5e3dd | 1876 | else if (ipv4_is_zeronet(fl4->daddr)) |
5ada5527 | 1877 | return ERR_PTR(-EINVAL); |
1da177e4 LT |
1878 | |
1879 | if (dev_out->flags & IFF_LOOPBACK) | |
1880 | flags |= RTCF_LOCAL; | |
1881 | ||
63617421 | 1882 | do_cache = true; |
982721f3 | 1883 | if (type == RTN_BROADCAST) { |
1da177e4 | 1884 | flags |= RTCF_BROADCAST | RTCF_LOCAL; |
982721f3 DM |
1885 | fi = NULL; |
1886 | } else if (type == RTN_MULTICAST) { | |
dd28d1a0 | 1887 | flags |= RTCF_MULTICAST | RTCF_LOCAL; |
813b3b5d DM |
1888 | if (!ip_check_mc_rcu(in_dev, fl4->daddr, fl4->saddr, |
1889 | fl4->flowi4_proto)) | |
1da177e4 | 1890 | flags &= ~RTCF_LOCAL; |
63617421 JA |
1891 | else |
1892 | do_cache = false; | |
1da177e4 | 1893 | /* If multicast route do not exist use |
dd28d1a0 ED |
1894 | * default one, but do not gateway in this case. |
1895 | * Yes, it is hack. | |
1da177e4 | 1896 | */ |
982721f3 DM |
1897 | if (fi && res->prefixlen < 4) |
1898 | fi = NULL; | |
1da177e4 LT |
1899 | } |
1900 | ||
f2bb4bed | 1901 | fnhe = NULL; |
63617421 JA |
1902 | do_cache &= fi != NULL; |
1903 | if (do_cache) { | |
c5038a83 | 1904 | struct rtable __rcu **prth; |
c92b9655 | 1905 | struct fib_nh *nh = &FIB_RES_NH(*res); |
d26b3a7c | 1906 | |
c92b9655 | 1907 | fnhe = find_exception(nh, fl4->daddr); |
c5038a83 | 1908 | if (fnhe) |
2ffae99d | 1909 | prth = &fnhe->fnhe_rth_output; |
c92b9655 JA |
1910 | else { |
1911 | if (unlikely(fl4->flowi4_flags & | |
1912 | FLOWI_FLAG_KNOWN_NH && | |
1913 | !(nh->nh_gw && | |
1914 | nh->nh_scope == RT_SCOPE_LINK))) { | |
1915 | do_cache = false; | |
1916 | goto add; | |
1917 | } | |
1918 | prth = __this_cpu_ptr(nh->nh_pcpu_rth_output); | |
1919 | } | |
c5038a83 DM |
1920 | rth = rcu_dereference(*prth); |
1921 | if (rt_cache_valid(rth)) { | |
1922 | dst_hold(&rth->dst); | |
1923 | return rth; | |
f2bb4bed DM |
1924 | } |
1925 | } | |
c92b9655 JA |
1926 | |
1927 | add: | |
5c1e6aa3 DM |
1928 | rth = rt_dst_alloc(dev_out, |
1929 | IN_DEV_CONF_GET(in_dev, NOPOLICY), | |
f2bb4bed | 1930 | IN_DEV_CONF_GET(in_dev, NOXFRM), |
c92b9655 | 1931 | do_cache); |
8391d07b | 1932 | if (!rth) |
5ada5527 | 1933 | return ERR_PTR(-ENOBUFS); |
8391d07b | 1934 | |
cf911662 DM |
1935 | rth->dst.output = ip_output; |
1936 | ||
ca4c3fc2 | 1937 | rth->rt_genid = rt_genid_ipv4(dev_net(dev_out)); |
cf911662 DM |
1938 | rth->rt_flags = flags; |
1939 | rth->rt_type = type; | |
9917e1e8 | 1940 | rth->rt_is_input = 0; |
13378cad | 1941 | rth->rt_iif = orig_oif ? : 0; |
5943634f | 1942 | rth->rt_pmtu = 0; |
f8126f1d | 1943 | rth->rt_gateway = 0; |
155e8336 | 1944 | rth->rt_uses_gateway = 0; |
caacf05e | 1945 | INIT_LIST_HEAD(&rth->rt_uncached); |
1da177e4 LT |
1946 | |
1947 | RT_CACHE_STAT_INC(out_slow_tot); | |
1948 | ||
41347dcd | 1949 | if (flags & RTCF_LOCAL) |
d8d1f30b | 1950 | rth->dst.input = ip_local_deliver; |
1da177e4 | 1951 | if (flags & (RTCF_BROADCAST | RTCF_MULTICAST)) { |
e905a9ed | 1952 | if (flags & RTCF_LOCAL && |
1da177e4 | 1953 | !(dev_out->flags & IFF_LOOPBACK)) { |
d8d1f30b | 1954 | rth->dst.output = ip_mc_output; |
1da177e4 LT |
1955 | RT_CACHE_STAT_INC(out_slow_mc); |
1956 | } | |
1957 | #ifdef CONFIG_IP_MROUTE | |
982721f3 | 1958 | if (type == RTN_MULTICAST) { |
1da177e4 | 1959 | if (IN_DEV_MFORWARD(in_dev) && |
813b3b5d | 1960 | !ipv4_is_local_multicast(fl4->daddr)) { |
d8d1f30b CG |
1961 | rth->dst.input = ip_mr_input; |
1962 | rth->dst.output = ip_mc_output; | |
1da177e4 LT |
1963 | } |
1964 | } | |
1965 | #endif | |
1966 | } | |
1967 | ||
f2bb4bed | 1968 | rt_set_nexthop(rth, fl4->daddr, res, fnhe, fi, type, 0); |
1da177e4 | 1969 | |
5ada5527 | 1970 | return rth; |
1da177e4 LT |
1971 | } |
1972 | ||
1da177e4 LT |
1973 | /* |
1974 | * Major route resolver routine. | |
1975 | */ | |
1976 | ||
89aef892 | 1977 | struct rtable *__ip_route_output_key(struct net *net, struct flowi4 *fl4) |
1da177e4 | 1978 | { |
1da177e4 | 1979 | struct net_device *dev_out = NULL; |
f61759e6 | 1980 | __u8 tos = RT_FL_TOS(fl4); |
813b3b5d DM |
1981 | unsigned int flags = 0; |
1982 | struct fib_result res; | |
5ada5527 | 1983 | struct rtable *rth; |
813b3b5d | 1984 | int orig_oif; |
1da177e4 | 1985 | |
85b91b03 | 1986 | res.tclassid = 0; |
1da177e4 | 1987 | res.fi = NULL; |
8b96d22d | 1988 | res.table = NULL; |
1da177e4 | 1989 | |
813b3b5d DM |
1990 | orig_oif = fl4->flowi4_oif; |
1991 | ||
1fb9489b | 1992 | fl4->flowi4_iif = LOOPBACK_IFINDEX; |
813b3b5d DM |
1993 | fl4->flowi4_tos = tos & IPTOS_RT_MASK; |
1994 | fl4->flowi4_scope = ((tos & RTO_ONLINK) ? | |
1995 | RT_SCOPE_LINK : RT_SCOPE_UNIVERSE); | |
44713b67 | 1996 | |
010c2708 | 1997 | rcu_read_lock(); |
813b3b5d | 1998 | if (fl4->saddr) { |
b23dd4fe | 1999 | rth = ERR_PTR(-EINVAL); |
813b3b5d DM |
2000 | if (ipv4_is_multicast(fl4->saddr) || |
2001 | ipv4_is_lbcast(fl4->saddr) || | |
2002 | ipv4_is_zeronet(fl4->saddr)) | |
1da177e4 LT |
2003 | goto out; |
2004 | ||
1da177e4 LT |
2005 | /* I removed check for oif == dev_out->oif here. |
2006 | It was wrong for two reasons: | |
1ab35276 DL |
2007 | 1. ip_dev_find(net, saddr) can return wrong iface, if saddr |
2008 | is assigned to multiple interfaces. | |
1da177e4 LT |
2009 | 2. Moreover, we are allowed to send packets with saddr |
2010 | of another iface. --ANK | |
2011 | */ | |
2012 | ||
813b3b5d DM |
2013 | if (fl4->flowi4_oif == 0 && |
2014 | (ipv4_is_multicast(fl4->daddr) || | |
2015 | ipv4_is_lbcast(fl4->daddr))) { | |
a210d01a | 2016 | /* It is equivalent to inet_addr_type(saddr) == RTN_LOCAL */ |
813b3b5d | 2017 | dev_out = __ip_dev_find(net, fl4->saddr, false); |
a210d01a JA |
2018 | if (dev_out == NULL) |
2019 | goto out; | |
2020 | ||
1da177e4 LT |
2021 | /* Special hack: user can direct multicasts |
2022 | and limited broadcast via necessary interface | |
2023 | without fiddling with IP_MULTICAST_IF or IP_PKTINFO. | |
2024 | This hack is not just for fun, it allows | |
2025 | vic,vat and friends to work. | |
2026 | They bind socket to loopback, set ttl to zero | |
2027 | and expect that it will work. | |
2028 | From the viewpoint of routing cache they are broken, | |
2029 | because we are not allowed to build multicast path | |
2030 | with loopback source addr (look, routing cache | |
2031 | cannot know, that ttl is zero, so that packet | |
2032 | will not leave this host and route is valid). | |
2033 | Luckily, this hack is good workaround. | |
2034 | */ | |
2035 | ||
813b3b5d | 2036 | fl4->flowi4_oif = dev_out->ifindex; |
1da177e4 LT |
2037 | goto make_route; |
2038 | } | |
a210d01a | 2039 | |
813b3b5d | 2040 | if (!(fl4->flowi4_flags & FLOWI_FLAG_ANYSRC)) { |
a210d01a | 2041 | /* It is equivalent to inet_addr_type(saddr) == RTN_LOCAL */ |
813b3b5d | 2042 | if (!__ip_dev_find(net, fl4->saddr, false)) |
a210d01a | 2043 | goto out; |
a210d01a | 2044 | } |
1da177e4 LT |
2045 | } |
2046 | ||
2047 | ||
813b3b5d DM |
2048 | if (fl4->flowi4_oif) { |
2049 | dev_out = dev_get_by_index_rcu(net, fl4->flowi4_oif); | |
b23dd4fe | 2050 | rth = ERR_PTR(-ENODEV); |
1da177e4 LT |
2051 | if (dev_out == NULL) |
2052 | goto out; | |
e5ed6399 HX |
2053 | |
2054 | /* RACE: Check return value of inet_select_addr instead. */ | |
fc75fc83 | 2055 | if (!(dev_out->flags & IFF_UP) || !__in_dev_get_rcu(dev_out)) { |
b23dd4fe | 2056 | rth = ERR_PTR(-ENETUNREACH); |
fc75fc83 ED |
2057 | goto out; |
2058 | } | |
813b3b5d DM |
2059 | if (ipv4_is_local_multicast(fl4->daddr) || |
2060 | ipv4_is_lbcast(fl4->daddr)) { | |
2061 | if (!fl4->saddr) | |
2062 | fl4->saddr = inet_select_addr(dev_out, 0, | |
2063 | RT_SCOPE_LINK); | |
1da177e4 LT |
2064 | goto make_route; |
2065 | } | |
0a7e2260 | 2066 | if (!fl4->saddr) { |
813b3b5d DM |
2067 | if (ipv4_is_multicast(fl4->daddr)) |
2068 | fl4->saddr = inet_select_addr(dev_out, 0, | |
2069 | fl4->flowi4_scope); | |
2070 | else if (!fl4->daddr) | |
2071 | fl4->saddr = inet_select_addr(dev_out, 0, | |
2072 | RT_SCOPE_HOST); | |
1da177e4 LT |
2073 | } |
2074 | } | |
2075 | ||
813b3b5d DM |
2076 | if (!fl4->daddr) { |
2077 | fl4->daddr = fl4->saddr; | |
2078 | if (!fl4->daddr) | |
2079 | fl4->daddr = fl4->saddr = htonl(INADDR_LOOPBACK); | |
b40afd0e | 2080 | dev_out = net->loopback_dev; |
1fb9489b | 2081 | fl4->flowi4_oif = LOOPBACK_IFINDEX; |
1da177e4 LT |
2082 | res.type = RTN_LOCAL; |
2083 | flags |= RTCF_LOCAL; | |
2084 | goto make_route; | |
2085 | } | |
2086 | ||
813b3b5d | 2087 | if (fib_lookup(net, fl4, &res)) { |
1da177e4 | 2088 | res.fi = NULL; |
8b96d22d | 2089 | res.table = NULL; |
813b3b5d | 2090 | if (fl4->flowi4_oif) { |
1da177e4 LT |
2091 | /* Apparently, routing tables are wrong. Assume, |
2092 | that the destination is on link. | |
2093 | ||
2094 | WHY? DW. | |
2095 | Because we are allowed to send to iface | |
2096 | even if it has NO routes and NO assigned | |
2097 | addresses. When oif is specified, routing | |
2098 | tables are looked up with only one purpose: | |
2099 | to catch if destination is gatewayed, rather than | |
2100 | direct. Moreover, if MSG_DONTROUTE is set, | |
2101 | we send packet, ignoring both routing tables | |
2102 | and ifaddr state. --ANK | |
2103 | ||
2104 | ||
2105 | We could make it even if oif is unknown, | |
2106 | likely IPv6, but we do not. | |
2107 | */ | |
2108 | ||
813b3b5d DM |
2109 | if (fl4->saddr == 0) |
2110 | fl4->saddr = inet_select_addr(dev_out, 0, | |
2111 | RT_SCOPE_LINK); | |
1da177e4 LT |
2112 | res.type = RTN_UNICAST; |
2113 | goto make_route; | |
2114 | } | |
b23dd4fe | 2115 | rth = ERR_PTR(-ENETUNREACH); |
1da177e4 LT |
2116 | goto out; |
2117 | } | |
1da177e4 LT |
2118 | |
2119 | if (res.type == RTN_LOCAL) { | |
813b3b5d | 2120 | if (!fl4->saddr) { |
9fc3bbb4 | 2121 | if (res.fi->fib_prefsrc) |
813b3b5d | 2122 | fl4->saddr = res.fi->fib_prefsrc; |
9fc3bbb4 | 2123 | else |
813b3b5d | 2124 | fl4->saddr = fl4->daddr; |
9fc3bbb4 | 2125 | } |
b40afd0e | 2126 | dev_out = net->loopback_dev; |
813b3b5d | 2127 | fl4->flowi4_oif = dev_out->ifindex; |
1da177e4 LT |
2128 | flags |= RTCF_LOCAL; |
2129 | goto make_route; | |
2130 | } | |
2131 | ||
2132 | #ifdef CONFIG_IP_ROUTE_MULTIPATH | |
813b3b5d | 2133 | if (res.fi->fib_nhs > 1 && fl4->flowi4_oif == 0) |
1b7fe593 | 2134 | fib_select_multipath(&res); |
1da177e4 LT |
2135 | else |
2136 | #endif | |
21d8c49e DM |
2137 | if (!res.prefixlen && |
2138 | res.table->tb_num_default > 1 && | |
813b3b5d | 2139 | res.type == RTN_UNICAST && !fl4->flowi4_oif) |
0c838ff1 | 2140 | fib_select_default(&res); |
1da177e4 | 2141 | |
813b3b5d DM |
2142 | if (!fl4->saddr) |
2143 | fl4->saddr = FIB_RES_PREFSRC(net, res); | |
1da177e4 | 2144 | |
1da177e4 | 2145 | dev_out = FIB_RES_DEV(res); |
813b3b5d | 2146 | fl4->flowi4_oif = dev_out->ifindex; |
1da177e4 LT |
2147 | |
2148 | ||
2149 | make_route: | |
1a00fee4 | 2150 | rth = __mkroute_output(&res, fl4, orig_oif, dev_out, flags); |
1da177e4 | 2151 | |
010c2708 DM |
2152 | out: |
2153 | rcu_read_unlock(); | |
b23dd4fe | 2154 | return rth; |
1da177e4 | 2155 | } |
d8c97a94 ACM |
2156 | EXPORT_SYMBOL_GPL(__ip_route_output_key); |
2157 | ||
ae2688d5 JW |
2158 | static struct dst_entry *ipv4_blackhole_dst_check(struct dst_entry *dst, u32 cookie) |
2159 | { | |
2160 | return NULL; | |
2161 | } | |
2162 | ||
ebb762f2 | 2163 | static unsigned int ipv4_blackhole_mtu(const struct dst_entry *dst) |
ec831ea7 | 2164 | { |
618f9bc7 SK |
2165 | unsigned int mtu = dst_metric_raw(dst, RTAX_MTU); |
2166 | ||
2167 | return mtu ? : dst->dev->mtu; | |
ec831ea7 RD |
2168 | } |
2169 | ||
6700c270 DM |
2170 | static void ipv4_rt_blackhole_update_pmtu(struct dst_entry *dst, struct sock *sk, |
2171 | struct sk_buff *skb, u32 mtu) | |
14e50e57 DM |
2172 | { |
2173 | } | |
2174 | ||
6700c270 DM |
2175 | static void ipv4_rt_blackhole_redirect(struct dst_entry *dst, struct sock *sk, |
2176 | struct sk_buff *skb) | |
b587ee3b DM |
2177 | { |
2178 | } | |
2179 | ||
0972ddb2 HB |
2180 | static u32 *ipv4_rt_blackhole_cow_metrics(struct dst_entry *dst, |
2181 | unsigned long old) | |
2182 | { | |
2183 | return NULL; | |
2184 | } | |
2185 | ||
14e50e57 DM |
2186 | static struct dst_ops ipv4_dst_blackhole_ops = { |
2187 | .family = AF_INET, | |
09640e63 | 2188 | .protocol = cpu_to_be16(ETH_P_IP), |
ae2688d5 | 2189 | .check = ipv4_blackhole_dst_check, |
ebb762f2 | 2190 | .mtu = ipv4_blackhole_mtu, |
214f45c9 | 2191 | .default_advmss = ipv4_default_advmss, |
14e50e57 | 2192 | .update_pmtu = ipv4_rt_blackhole_update_pmtu, |
b587ee3b | 2193 | .redirect = ipv4_rt_blackhole_redirect, |
0972ddb2 | 2194 | .cow_metrics = ipv4_rt_blackhole_cow_metrics, |
d3aaeb38 | 2195 | .neigh_lookup = ipv4_neigh_lookup, |
14e50e57 DM |
2196 | }; |
2197 | ||
2774c131 | 2198 | struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_orig) |
14e50e57 | 2199 | { |
2774c131 | 2200 | struct rtable *ort = (struct rtable *) dst_orig; |
f5b0a874 | 2201 | struct rtable *rt; |
14e50e57 | 2202 | |
f5b0a874 | 2203 | rt = dst_alloc(&ipv4_dst_blackhole_ops, NULL, 1, DST_OBSOLETE_NONE, 0); |
14e50e57 | 2204 | if (rt) { |
d8d1f30b | 2205 | struct dst_entry *new = &rt->dst; |
14e50e57 | 2206 | |
14e50e57 | 2207 | new->__use = 1; |
352e512c | 2208 | new->input = dst_discard; |
aad88724 | 2209 | new->output = dst_discard_sk; |
14e50e57 | 2210 | |
d8d1f30b | 2211 | new->dev = ort->dst.dev; |
14e50e57 DM |
2212 | if (new->dev) |
2213 | dev_hold(new->dev); | |
2214 | ||
9917e1e8 | 2215 | rt->rt_is_input = ort->rt_is_input; |
5e2b61f7 | 2216 | rt->rt_iif = ort->rt_iif; |
5943634f | 2217 | rt->rt_pmtu = ort->rt_pmtu; |
14e50e57 | 2218 | |
ca4c3fc2 | 2219 | rt->rt_genid = rt_genid_ipv4(net); |
14e50e57 DM |
2220 | rt->rt_flags = ort->rt_flags; |
2221 | rt->rt_type = ort->rt_type; | |
14e50e57 | 2222 | rt->rt_gateway = ort->rt_gateway; |
155e8336 | 2223 | rt->rt_uses_gateway = ort->rt_uses_gateway; |
14e50e57 | 2224 | |
caacf05e DM |
2225 | INIT_LIST_HEAD(&rt->rt_uncached); |
2226 | ||
14e50e57 DM |
2227 | dst_free(new); |
2228 | } | |
2229 | ||
2774c131 DM |
2230 | dst_release(dst_orig); |
2231 | ||
2232 | return rt ? &rt->dst : ERR_PTR(-ENOMEM); | |
14e50e57 DM |
2233 | } |
2234 | ||
9d6ec938 | 2235 | struct rtable *ip_route_output_flow(struct net *net, struct flowi4 *flp4, |
b23dd4fe | 2236 | struct sock *sk) |
1da177e4 | 2237 | { |
9d6ec938 | 2238 | struct rtable *rt = __ip_route_output_key(net, flp4); |
1da177e4 | 2239 | |
b23dd4fe DM |
2240 | if (IS_ERR(rt)) |
2241 | return rt; | |
1da177e4 | 2242 | |
56157872 | 2243 | if (flp4->flowi4_proto) |
9d6ec938 DM |
2244 | rt = (struct rtable *) xfrm_lookup(net, &rt->dst, |
2245 | flowi4_to_flowi(flp4), | |
2246 | sk, 0); | |
1da177e4 | 2247 | |
b23dd4fe | 2248 | return rt; |
1da177e4 | 2249 | } |
d8c97a94 ACM |
2250 | EXPORT_SYMBOL_GPL(ip_route_output_flow); |
2251 | ||
f1ce3062 | 2252 | static int rt_fill_info(struct net *net, __be32 dst, __be32 src, |
15e47304 | 2253 | struct flowi4 *fl4, struct sk_buff *skb, u32 portid, |
f1ce3062 | 2254 | u32 seq, int event, int nowait, unsigned int flags) |
1da177e4 | 2255 | { |
511c3f92 | 2256 | struct rtable *rt = skb_rtable(skb); |
1da177e4 | 2257 | struct rtmsg *r; |
be403ea1 | 2258 | struct nlmsghdr *nlh; |
2bc8ca40 | 2259 | unsigned long expires = 0; |
f185071d | 2260 | u32 error; |
521f5490 | 2261 | u32 metrics[RTAX_MAX]; |
be403ea1 | 2262 | |
15e47304 | 2263 | nlh = nlmsg_put(skb, portid, seq, event, sizeof(*r), flags); |
be403ea1 | 2264 | if (nlh == NULL) |
26932566 | 2265 | return -EMSGSIZE; |
be403ea1 TG |
2266 | |
2267 | r = nlmsg_data(nlh); | |
1da177e4 LT |
2268 | r->rtm_family = AF_INET; |
2269 | r->rtm_dst_len = 32; | |
2270 | r->rtm_src_len = 0; | |
d6c0a4f6 | 2271 | r->rtm_tos = fl4->flowi4_tos; |
1da177e4 | 2272 | r->rtm_table = RT_TABLE_MAIN; |
f3756b79 DM |
2273 | if (nla_put_u32(skb, RTA_TABLE, RT_TABLE_MAIN)) |
2274 | goto nla_put_failure; | |
1da177e4 LT |
2275 | r->rtm_type = rt->rt_type; |
2276 | r->rtm_scope = RT_SCOPE_UNIVERSE; | |
2277 | r->rtm_protocol = RTPROT_UNSPEC; | |
2278 | r->rtm_flags = (rt->rt_flags & ~0xFFFF) | RTM_F_CLONED; | |
2279 | if (rt->rt_flags & RTCF_NOTIFY) | |
2280 | r->rtm_flags |= RTM_F_NOTIFY; | |
be403ea1 | 2281 | |
f1ce3062 | 2282 | if (nla_put_be32(skb, RTA_DST, dst)) |
f3756b79 | 2283 | goto nla_put_failure; |
1a00fee4 | 2284 | if (src) { |
1da177e4 | 2285 | r->rtm_src_len = 32; |
1a00fee4 | 2286 | if (nla_put_be32(skb, RTA_SRC, src)) |
f3756b79 | 2287 | goto nla_put_failure; |
1da177e4 | 2288 | } |
f3756b79 DM |
2289 | if (rt->dst.dev && |
2290 | nla_put_u32(skb, RTA_OIF, rt->dst.dev->ifindex)) | |
2291 | goto nla_put_failure; | |
c7066f70 | 2292 | #ifdef CONFIG_IP_ROUTE_CLASSID |
f3756b79 DM |
2293 | if (rt->dst.tclassid && |
2294 | nla_put_u32(skb, RTA_FLOW, rt->dst.tclassid)) | |
2295 | goto nla_put_failure; | |
1da177e4 | 2296 | #endif |
41347dcd | 2297 | if (!rt_is_input_route(rt) && |
d6c0a4f6 DM |
2298 | fl4->saddr != src) { |
2299 | if (nla_put_be32(skb, RTA_PREFSRC, fl4->saddr)) | |
f3756b79 DM |
2300 | goto nla_put_failure; |
2301 | } | |
155e8336 | 2302 | if (rt->rt_uses_gateway && |
f3756b79 DM |
2303 | nla_put_be32(skb, RTA_GATEWAY, rt->rt_gateway)) |
2304 | goto nla_put_failure; | |
be403ea1 | 2305 | |
ee9a8f7a SK |
2306 | expires = rt->dst.expires; |
2307 | if (expires) { | |
2308 | unsigned long now = jiffies; | |
2309 | ||
2310 | if (time_before(now, expires)) | |
2311 | expires -= now; | |
2312 | else | |
2313 | expires = 0; | |
2314 | } | |
2315 | ||
521f5490 | 2316 | memcpy(metrics, dst_metrics_ptr(&rt->dst), sizeof(metrics)); |
ee9a8f7a | 2317 | if (rt->rt_pmtu && expires) |
521f5490 JA |
2318 | metrics[RTAX_MTU - 1] = rt->rt_pmtu; |
2319 | if (rtnetlink_put_metrics(skb, metrics) < 0) | |
be403ea1 TG |
2320 | goto nla_put_failure; |
2321 | ||
b4869889 | 2322 | if (fl4->flowi4_mark && |
68aaed54 | 2323 | nla_put_u32(skb, RTA_MARK, fl4->flowi4_mark)) |
f3756b79 | 2324 | goto nla_put_failure; |
963bfeee | 2325 | |
d8d1f30b | 2326 | error = rt->dst.error; |
be403ea1 | 2327 | |
c7537967 | 2328 | if (rt_is_input_route(rt)) { |
8caaf7b6 ND |
2329 | #ifdef CONFIG_IP_MROUTE |
2330 | if (ipv4_is_multicast(dst) && !ipv4_is_local_multicast(dst) && | |
2331 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)) { | |
2332 | int err = ipmr_get_route(net, skb, | |
2333 | fl4->saddr, fl4->daddr, | |
2334 | r, nowait); | |
2335 | if (err <= 0) { | |
2336 | if (!nowait) { | |
2337 | if (err == 0) | |
2338 | return 0; | |
2339 | goto nla_put_failure; | |
2340 | } else { | |
2341 | if (err == -EMSGSIZE) | |
2342 | goto nla_put_failure; | |
2343 | error = err; | |
2344 | } | |
2345 | } | |
2346 | } else | |
2347 | #endif | |
91146153 | 2348 | if (nla_put_u32(skb, RTA_IIF, skb->dev->ifindex)) |
8caaf7b6 | 2349 | goto nla_put_failure; |
1da177e4 LT |
2350 | } |
2351 | ||
f185071d | 2352 | if (rtnl_put_cacheinfo(skb, &rt->dst, 0, expires, error) < 0) |
e3703b3d | 2353 | goto nla_put_failure; |
be403ea1 TG |
2354 | |
2355 | return nlmsg_end(skb, nlh); | |
1da177e4 | 2356 | |
be403ea1 | 2357 | nla_put_failure: |
26932566 PM |
2358 | nlmsg_cancel(skb, nlh); |
2359 | return -EMSGSIZE; | |
1da177e4 LT |
2360 | } |
2361 | ||
661d2967 | 2362 | static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh) |
1da177e4 | 2363 | { |
3b1e0a65 | 2364 | struct net *net = sock_net(in_skb->sk); |
d889ce3b TG |
2365 | struct rtmsg *rtm; |
2366 | struct nlattr *tb[RTA_MAX+1]; | |
1da177e4 | 2367 | struct rtable *rt = NULL; |
d6c0a4f6 | 2368 | struct flowi4 fl4; |
9e12bb22 AV |
2369 | __be32 dst = 0; |
2370 | __be32 src = 0; | |
2371 | u32 iif; | |
d889ce3b | 2372 | int err; |
963bfeee | 2373 | int mark; |
1da177e4 LT |
2374 | struct sk_buff *skb; |
2375 | ||
d889ce3b TG |
2376 | err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv4_policy); |
2377 | if (err < 0) | |
2378 | goto errout; | |
2379 | ||
2380 | rtm = nlmsg_data(nlh); | |
2381 | ||
1da177e4 | 2382 | skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); |
d889ce3b TG |
2383 | if (skb == NULL) { |
2384 | err = -ENOBUFS; | |
2385 | goto errout; | |
2386 | } | |
1da177e4 LT |
2387 | |
2388 | /* Reserve room for dummy headers, this skb can pass | |
2389 | through good chunk of routing engine. | |
2390 | */ | |
459a98ed | 2391 | skb_reset_mac_header(skb); |
c1d2bbe1 | 2392 | skb_reset_network_header(skb); |
d2c962b8 SH |
2393 | |
2394 | /* Bugfix: need to give ip_route_input enough of an IP header to not gag. */ | |
eddc9ec5 | 2395 | ip_hdr(skb)->protocol = IPPROTO_ICMP; |
1da177e4 LT |
2396 | skb_reserve(skb, MAX_HEADER + sizeof(struct iphdr)); |
2397 | ||
17fb2c64 AV |
2398 | src = tb[RTA_SRC] ? nla_get_be32(tb[RTA_SRC]) : 0; |
2399 | dst = tb[RTA_DST] ? nla_get_be32(tb[RTA_DST]) : 0; | |
d889ce3b | 2400 | iif = tb[RTA_IIF] ? nla_get_u32(tb[RTA_IIF]) : 0; |
963bfeee | 2401 | mark = tb[RTA_MARK] ? nla_get_u32(tb[RTA_MARK]) : 0; |
1da177e4 | 2402 | |
d6c0a4f6 DM |
2403 | memset(&fl4, 0, sizeof(fl4)); |
2404 | fl4.daddr = dst; | |
2405 | fl4.saddr = src; | |
2406 | fl4.flowi4_tos = rtm->rtm_tos; | |
2407 | fl4.flowi4_oif = tb[RTA_OIF] ? nla_get_u32(tb[RTA_OIF]) : 0; | |
2408 | fl4.flowi4_mark = mark; | |
2409 | ||
1da177e4 | 2410 | if (iif) { |
d889ce3b TG |
2411 | struct net_device *dev; |
2412 | ||
1937504d | 2413 | dev = __dev_get_by_index(net, iif); |
d889ce3b TG |
2414 | if (dev == NULL) { |
2415 | err = -ENODEV; | |
2416 | goto errout_free; | |
2417 | } | |
2418 | ||
1da177e4 LT |
2419 | skb->protocol = htons(ETH_P_IP); |
2420 | skb->dev = dev; | |
963bfeee | 2421 | skb->mark = mark; |
1da177e4 LT |
2422 | local_bh_disable(); |
2423 | err = ip_route_input(skb, dst, src, rtm->rtm_tos, dev); | |
2424 | local_bh_enable(); | |
d889ce3b | 2425 | |
511c3f92 | 2426 | rt = skb_rtable(skb); |
d8d1f30b CG |
2427 | if (err == 0 && rt->dst.error) |
2428 | err = -rt->dst.error; | |
1da177e4 | 2429 | } else { |
9d6ec938 | 2430 | rt = ip_route_output_key(net, &fl4); |
b23dd4fe DM |
2431 | |
2432 | err = 0; | |
2433 | if (IS_ERR(rt)) | |
2434 | err = PTR_ERR(rt); | |
1da177e4 | 2435 | } |
d889ce3b | 2436 | |
1da177e4 | 2437 | if (err) |
d889ce3b | 2438 | goto errout_free; |
1da177e4 | 2439 | |
d8d1f30b | 2440 | skb_dst_set(skb, &rt->dst); |
1da177e4 LT |
2441 | if (rtm->rtm_flags & RTM_F_NOTIFY) |
2442 | rt->rt_flags |= RTCF_NOTIFY; | |
2443 | ||
f1ce3062 | 2444 | err = rt_fill_info(net, dst, src, &fl4, skb, |
15e47304 | 2445 | NETLINK_CB(in_skb).portid, nlh->nlmsg_seq, |
1937504d | 2446 | RTM_NEWROUTE, 0, 0); |
d889ce3b TG |
2447 | if (err <= 0) |
2448 | goto errout_free; | |
1da177e4 | 2449 | |
15e47304 | 2450 | err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid); |
d889ce3b | 2451 | errout: |
2942e900 | 2452 | return err; |
1da177e4 | 2453 | |
d889ce3b | 2454 | errout_free: |
1da177e4 | 2455 | kfree_skb(skb); |
d889ce3b | 2456 | goto errout; |
1da177e4 LT |
2457 | } |
2458 | ||
1da177e4 LT |
2459 | void ip_rt_multicast_event(struct in_device *in_dev) |
2460 | { | |
4ccfe6d4 | 2461 | rt_cache_flush(dev_net(in_dev->dev)); |
1da177e4 LT |
2462 | } |
2463 | ||
2464 | #ifdef CONFIG_SYSCTL | |
082c7ca4 G |
2465 | static int ip_rt_gc_timeout __read_mostly = RT_GC_TIMEOUT; |
2466 | static int ip_rt_gc_interval __read_mostly = 60 * HZ; | |
2467 | static int ip_rt_gc_min_interval __read_mostly = HZ / 2; | |
2468 | static int ip_rt_gc_elasticity __read_mostly = 8; | |
2469 | ||
fe2c6338 | 2470 | static int ipv4_sysctl_rtcache_flush(struct ctl_table *__ctl, int write, |
8d65af78 | 2471 | void __user *buffer, |
1da177e4 LT |
2472 | size_t *lenp, loff_t *ppos) |
2473 | { | |
5aad1de5 TT |
2474 | struct net *net = (struct net *)__ctl->extra1; |
2475 | ||
1da177e4 | 2476 | if (write) { |
5aad1de5 TT |
2477 | rt_cache_flush(net); |
2478 | fnhe_genid_bump(net); | |
1da177e4 | 2479 | return 0; |
e905a9ed | 2480 | } |
1da177e4 LT |
2481 | |
2482 | return -EINVAL; | |
2483 | } | |
2484 | ||
fe2c6338 | 2485 | static struct ctl_table ipv4_route_table[] = { |
1da177e4 | 2486 | { |
1da177e4 LT |
2487 | .procname = "gc_thresh", |
2488 | .data = &ipv4_dst_ops.gc_thresh, | |
2489 | .maxlen = sizeof(int), | |
2490 | .mode = 0644, | |
6d9f239a | 2491 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2492 | }, |
2493 | { | |
1da177e4 LT |
2494 | .procname = "max_size", |
2495 | .data = &ip_rt_max_size, | |
2496 | .maxlen = sizeof(int), | |
2497 | .mode = 0644, | |
6d9f239a | 2498 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2499 | }, |
2500 | { | |
2501 | /* Deprecated. Use gc_min_interval_ms */ | |
e905a9ed | 2502 | |
1da177e4 LT |
2503 | .procname = "gc_min_interval", |
2504 | .data = &ip_rt_gc_min_interval, | |
2505 | .maxlen = sizeof(int), | |
2506 | .mode = 0644, | |
6d9f239a | 2507 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2508 | }, |
2509 | { | |
1da177e4 LT |
2510 | .procname = "gc_min_interval_ms", |
2511 | .data = &ip_rt_gc_min_interval, | |
2512 | .maxlen = sizeof(int), | |
2513 | .mode = 0644, | |
6d9f239a | 2514 | .proc_handler = proc_dointvec_ms_jiffies, |
1da177e4 LT |
2515 | }, |
2516 | { | |
1da177e4 LT |
2517 | .procname = "gc_timeout", |
2518 | .data = &ip_rt_gc_timeout, | |
2519 | .maxlen = sizeof(int), | |
2520 | .mode = 0644, | |
6d9f239a | 2521 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 | 2522 | }, |
9f28a2fc ED |
2523 | { |
2524 | .procname = "gc_interval", | |
2525 | .data = &ip_rt_gc_interval, | |
2526 | .maxlen = sizeof(int), | |
2527 | .mode = 0644, | |
2528 | .proc_handler = proc_dointvec_jiffies, | |
2529 | }, | |
1da177e4 | 2530 | { |
1da177e4 LT |
2531 | .procname = "redirect_load", |
2532 | .data = &ip_rt_redirect_load, | |
2533 | .maxlen = sizeof(int), | |
2534 | .mode = 0644, | |
6d9f239a | 2535 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2536 | }, |
2537 | { | |
1da177e4 LT |
2538 | .procname = "redirect_number", |
2539 | .data = &ip_rt_redirect_number, | |
2540 | .maxlen = sizeof(int), | |
2541 | .mode = 0644, | |
6d9f239a | 2542 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2543 | }, |
2544 | { | |
1da177e4 LT |
2545 | .procname = "redirect_silence", |
2546 | .data = &ip_rt_redirect_silence, | |
2547 | .maxlen = sizeof(int), | |
2548 | .mode = 0644, | |
6d9f239a | 2549 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2550 | }, |
2551 | { | |
1da177e4 LT |
2552 | .procname = "error_cost", |
2553 | .data = &ip_rt_error_cost, | |
2554 | .maxlen = sizeof(int), | |
2555 | .mode = 0644, | |
6d9f239a | 2556 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2557 | }, |
2558 | { | |
1da177e4 LT |
2559 | .procname = "error_burst", |
2560 | .data = &ip_rt_error_burst, | |
2561 | .maxlen = sizeof(int), | |
2562 | .mode = 0644, | |
6d9f239a | 2563 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2564 | }, |
2565 | { | |
1da177e4 LT |
2566 | .procname = "gc_elasticity", |
2567 | .data = &ip_rt_gc_elasticity, | |
2568 | .maxlen = sizeof(int), | |
2569 | .mode = 0644, | |
6d9f239a | 2570 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2571 | }, |
2572 | { | |
1da177e4 LT |
2573 | .procname = "mtu_expires", |
2574 | .data = &ip_rt_mtu_expires, | |
2575 | .maxlen = sizeof(int), | |
2576 | .mode = 0644, | |
6d9f239a | 2577 | .proc_handler = proc_dointvec_jiffies, |
1da177e4 LT |
2578 | }, |
2579 | { | |
1da177e4 LT |
2580 | .procname = "min_pmtu", |
2581 | .data = &ip_rt_min_pmtu, | |
2582 | .maxlen = sizeof(int), | |
2583 | .mode = 0644, | |
6d9f239a | 2584 | .proc_handler = proc_dointvec, |
1da177e4 LT |
2585 | }, |
2586 | { | |
1da177e4 LT |
2587 | .procname = "min_adv_mss", |
2588 | .data = &ip_rt_min_advmss, | |
2589 | .maxlen = sizeof(int), | |
2590 | .mode = 0644, | |
6d9f239a | 2591 | .proc_handler = proc_dointvec, |
1da177e4 | 2592 | }, |
f8572d8f | 2593 | { } |
1da177e4 | 2594 | }; |
39a23e75 | 2595 | |
39a23e75 DL |
2596 | static struct ctl_table ipv4_route_flush_table[] = { |
2597 | { | |
39a23e75 DL |
2598 | .procname = "flush", |
2599 | .maxlen = sizeof(int), | |
2600 | .mode = 0200, | |
6d9f239a | 2601 | .proc_handler = ipv4_sysctl_rtcache_flush, |
39a23e75 | 2602 | }, |
f8572d8f | 2603 | { }, |
39a23e75 DL |
2604 | }; |
2605 | ||
2606 | static __net_init int sysctl_route_net_init(struct net *net) | |
2607 | { | |
2608 | struct ctl_table *tbl; | |
2609 | ||
2610 | tbl = ipv4_route_flush_table; | |
09ad9bc7 | 2611 | if (!net_eq(net, &init_net)) { |
39a23e75 DL |
2612 | tbl = kmemdup(tbl, sizeof(ipv4_route_flush_table), GFP_KERNEL); |
2613 | if (tbl == NULL) | |
2614 | goto err_dup; | |
464dc801 EB |
2615 | |
2616 | /* Don't export sysctls to unprivileged users */ | |
2617 | if (net->user_ns != &init_user_ns) | |
2618 | tbl[0].procname = NULL; | |
39a23e75 DL |
2619 | } |
2620 | tbl[0].extra1 = net; | |
2621 | ||
ec8f23ce | 2622 | net->ipv4.route_hdr = register_net_sysctl(net, "net/ipv4/route", tbl); |
39a23e75 DL |
2623 | if (net->ipv4.route_hdr == NULL) |
2624 | goto err_reg; | |
2625 | return 0; | |
2626 | ||
2627 | err_reg: | |
2628 | if (tbl != ipv4_route_flush_table) | |
2629 | kfree(tbl); | |
2630 | err_dup: | |
2631 | return -ENOMEM; | |
2632 | } | |
2633 | ||
2634 | static __net_exit void sysctl_route_net_exit(struct net *net) | |
2635 | { | |
2636 | struct ctl_table *tbl; | |
2637 | ||
2638 | tbl = net->ipv4.route_hdr->ctl_table_arg; | |
2639 | unregister_net_sysctl_table(net->ipv4.route_hdr); | |
2640 | BUG_ON(tbl == ipv4_route_flush_table); | |
2641 | kfree(tbl); | |
2642 | } | |
2643 | ||
2644 | static __net_initdata struct pernet_operations sysctl_route_ops = { | |
2645 | .init = sysctl_route_net_init, | |
2646 | .exit = sysctl_route_net_exit, | |
2647 | }; | |
1da177e4 LT |
2648 | #endif |
2649 | ||
3ee94372 | 2650 | static __net_init int rt_genid_init(struct net *net) |
9f5e97e5 | 2651 | { |
ca4c3fc2 | 2652 | atomic_set(&net->ipv4.rt_genid, 0); |
5aad1de5 | 2653 | atomic_set(&net->fnhe_genid, 0); |
436c3b66 DM |
2654 | get_random_bytes(&net->ipv4.dev_addr_genid, |
2655 | sizeof(net->ipv4.dev_addr_genid)); | |
9f5e97e5 DL |
2656 | return 0; |
2657 | } | |
2658 | ||
3ee94372 NH |
2659 | static __net_initdata struct pernet_operations rt_genid_ops = { |
2660 | .init = rt_genid_init, | |
9f5e97e5 DL |
2661 | }; |
2662 | ||
c3426b47 DM |
2663 | static int __net_init ipv4_inetpeer_init(struct net *net) |
2664 | { | |
2665 | struct inet_peer_base *bp = kmalloc(sizeof(*bp), GFP_KERNEL); | |
2666 | ||
2667 | if (!bp) | |
2668 | return -ENOMEM; | |
2669 | inet_peer_base_init(bp); | |
2670 | net->ipv4.peers = bp; | |
2671 | return 0; | |
2672 | } | |
2673 | ||
2674 | static void __net_exit ipv4_inetpeer_exit(struct net *net) | |
2675 | { | |
2676 | struct inet_peer_base *bp = net->ipv4.peers; | |
2677 | ||
2678 | net->ipv4.peers = NULL; | |
56a6b248 | 2679 | inetpeer_invalidate_tree(bp); |
c3426b47 DM |
2680 | kfree(bp); |
2681 | } | |
2682 | ||
2683 | static __net_initdata struct pernet_operations ipv4_inetpeer_ops = { | |
2684 | .init = ipv4_inetpeer_init, | |
2685 | .exit = ipv4_inetpeer_exit, | |
2686 | }; | |
9f5e97e5 | 2687 | |
c7066f70 | 2688 | #ifdef CONFIG_IP_ROUTE_CLASSID |
7d720c3e | 2689 | struct ip_rt_acct __percpu *ip_rt_acct __read_mostly; |
c7066f70 | 2690 | #endif /* CONFIG_IP_ROUTE_CLASSID */ |
1da177e4 | 2691 | |
1da177e4 LT |
2692 | int __init ip_rt_init(void) |
2693 | { | |
424c4b70 | 2694 | int rc = 0; |
1da177e4 | 2695 | |
73f156a6 ED |
2696 | ip_idents = kmalloc(IP_IDENTS_SZ * sizeof(*ip_idents), GFP_KERNEL); |
2697 | if (!ip_idents) | |
2698 | panic("IP: failed to allocate ip_idents\n"); | |
2699 | ||
2700 | prandom_bytes(ip_idents, IP_IDENTS_SZ * sizeof(*ip_idents)); | |
2701 | ||
c7066f70 | 2702 | #ifdef CONFIG_IP_ROUTE_CLASSID |
0dcec8c2 | 2703 | ip_rt_acct = __alloc_percpu(256 * sizeof(struct ip_rt_acct), __alignof__(struct ip_rt_acct)); |
1da177e4 LT |
2704 | if (!ip_rt_acct) |
2705 | panic("IP: failed to allocate ip_rt_acct\n"); | |
1da177e4 LT |
2706 | #endif |
2707 | ||
e5d679f3 AD |
2708 | ipv4_dst_ops.kmem_cachep = |
2709 | kmem_cache_create("ip_dst_cache", sizeof(struct rtable), 0, | |
20c2df83 | 2710 | SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL); |
1da177e4 | 2711 | |
14e50e57 DM |
2712 | ipv4_dst_blackhole_ops.kmem_cachep = ipv4_dst_ops.kmem_cachep; |
2713 | ||
fc66f95c ED |
2714 | if (dst_entries_init(&ipv4_dst_ops) < 0) |
2715 | panic("IP: failed to allocate ipv4_dst_ops counter\n"); | |
2716 | ||
2717 | if (dst_entries_init(&ipv4_dst_blackhole_ops) < 0) | |
2718 | panic("IP: failed to allocate ipv4_dst_blackhole_ops counter\n"); | |
2719 | ||
89aef892 DM |
2720 | ipv4_dst_ops.gc_thresh = ~0; |
2721 | ip_rt_max_size = INT_MAX; | |
1da177e4 | 2722 | |
1da177e4 LT |
2723 | devinet_init(); |
2724 | ip_fib_init(); | |
2725 | ||
73b38711 | 2726 | if (ip_rt_proc_init()) |
058bd4d2 | 2727 | pr_err("Unable to create route proc files\n"); |
1da177e4 LT |
2728 | #ifdef CONFIG_XFRM |
2729 | xfrm_init(); | |
703fb94e | 2730 | xfrm4_init(); |
1da177e4 | 2731 | #endif |
c7ac8679 | 2732 | rtnl_register(PF_INET, RTM_GETROUTE, inet_rtm_getroute, NULL, NULL); |
63f3444f | 2733 | |
39a23e75 DL |
2734 | #ifdef CONFIG_SYSCTL |
2735 | register_pernet_subsys(&sysctl_route_ops); | |
2736 | #endif | |
3ee94372 | 2737 | register_pernet_subsys(&rt_genid_ops); |
c3426b47 | 2738 | register_pernet_subsys(&ipv4_inetpeer_ops); |
1da177e4 LT |
2739 | return rc; |
2740 | } | |
2741 | ||
a1bc6eb4 | 2742 | #ifdef CONFIG_SYSCTL |
eeb61f71 AV |
2743 | /* |
2744 | * We really need to sanitize the damn ipv4 init order, then all | |
2745 | * this nonsense will go away. | |
2746 | */ | |
2747 | void __init ip_static_sysctl_init(void) | |
2748 | { | |
4e5ca785 | 2749 | register_net_sysctl(&init_net, "net/ipv4/route", ipv4_route_table); |
eeb61f71 | 2750 | } |
a1bc6eb4 | 2751 | #endif |