1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16#include <asm/uaccess.h>
17#include <asm/system.h>
18#include <linux/bitops.h>
19#include <linux/types.h>
20#include <linux/kernel.h>
21#include <linux/jiffies.h>
22#include <linux/mm.h>
23#include <linux/string.h>
24#include <linux/socket.h>
25#include <linux/sockios.h>
26#include <linux/errno.h>
27#include <linux/in.h>
28#include <linux/inet.h>
29#include <linux/inetdevice.h>
30#include <linux/netdevice.h>
31#include <linux/if_arp.h>
32#include <linux/proc_fs.h>
33#include <linux/skbuff.h>
34#include <linux/init.h>
35
36#include <net/arp.h>
37#include <net/ip.h>
38#include <net/protocol.h>
39#include <net/route.h>
40#include <net/tcp.h>
41#include <net/sock.h>
42#include <net/ip_fib.h>
43#include <net/netlink.h>
44#include <net/nexthop.h>
45
46#include "fib_lookup.h"
47
48static DEFINE_SPINLOCK(fib_info_lock);
49static struct hlist_head *fib_info_hash;
50static struct hlist_head *fib_info_laddrhash;
51static unsigned int fib_hash_size;
52static unsigned int fib_info_cnt;
53
54#define DEVINDEX_HASHBITS 8
55#define DEVINDEX_HASHSIZE (1U << DEVINDEX_HASHBITS)
56static struct hlist_head fib_info_devhash[DEVINDEX_HASHSIZE];
57
58#ifdef CONFIG_IP_ROUTE_MULTIPATH
59
60static DEFINE_SPINLOCK(fib_multipath_lock);
61
62#define for_nexthops(fi) { int nhsel; const struct fib_nh * nh; \
63for (nhsel=0, nh = (fi)->fib_nh; nhsel < (fi)->fib_nhs; nh++, nhsel++)
64
65#define change_nexthops(fi) { int nhsel; struct fib_nh * nh; \
66for (nhsel=0, nh = (struct fib_nh *)((fi)->fib_nh); nhsel < (fi)->fib_nhs; nh++, nhsel++)
67
68#else
69
70
71
72#define for_nexthops(fi) { int nhsel = 0; const struct fib_nh * nh = (fi)->fib_nh; \
73for (nhsel=0; nhsel < 1; nhsel++)
74
75#define change_nexthops(fi) { int nhsel = 0; struct fib_nh * nh = (struct fib_nh *)((fi)->fib_nh); \
76for (nhsel=0; nhsel < 1; nhsel++)
77
78#endif
79
80#define endfor_nexthops(fi) }
81
82
83static const struct
84{
85 int error;
86 u8 scope;
87} fib_props[RTN_MAX + 1] = {
88 {
89 .error = 0,
90 .scope = RT_SCOPE_NOWHERE,
91 },
92 {
93 .error = 0,
94 .scope = RT_SCOPE_UNIVERSE,
95 },
96 {
97 .error = 0,
98 .scope = RT_SCOPE_HOST,
99 },
100 {
101 .error = 0,
102 .scope = RT_SCOPE_LINK,
103 },
104 {
105 .error = 0,
106 .scope = RT_SCOPE_LINK,
107 },
108 {
109 .error = 0,
110 .scope = RT_SCOPE_UNIVERSE,
111 },
112 {
113 .error = -EINVAL,
114 .scope = RT_SCOPE_UNIVERSE,
115 },
116 {
117 .error = -EHOSTUNREACH,
118 .scope = RT_SCOPE_UNIVERSE,
119 },
120 {
121 .error = -EACCES,
122 .scope = RT_SCOPE_UNIVERSE,
123 },
124 {
125 .error = -EAGAIN,
126 .scope = RT_SCOPE_UNIVERSE,
127 },
128 {
129 .error = -EINVAL,
130 .scope = RT_SCOPE_NOWHERE,
131 },
132 {
133 .error = -EINVAL,
134 .scope = RT_SCOPE_NOWHERE,
135 },
136};
137
138
139
140
141void free_fib_info(struct fib_info *fi)
142{
143 if (fi->fib_dead == 0) {
144 printk(KERN_WARNING "Freeing alive fib_info %p\n", fi);
145 return;
146 }
147 change_nexthops(fi) {
148 if (nh->nh_dev)
149 dev_put(nh->nh_dev);
150 nh->nh_dev = NULL;
151 } endfor_nexthops(fi);
152 fib_info_cnt--;
153 release_net(fi->fib_net);
154 kfree(fi);
155}
156
157void fib_release_info(struct fib_info *fi)
158{
159 spin_lock_bh(&fib_info_lock);
160 if (fi && --fi->fib_treeref == 0) {
161 hlist_del(&fi->fib_hash);
162 if (fi->fib_prefsrc)
163 hlist_del(&fi->fib_lhash);
164 change_nexthops(fi) {
165 if (!nh->nh_dev)
166 continue;
167 hlist_del(&nh->nh_hash);
168 } endfor_nexthops(fi)
169 fi->fib_dead = 1;
170 fib_info_put(fi);
171 }
172 spin_unlock_bh(&fib_info_lock);
173}
174
175static __inline__ int nh_comp(const struct fib_info *fi, const struct fib_info *ofi)
176{
177 const struct fib_nh *onh = ofi->fib_nh;
178
179 for_nexthops(fi) {
180 if (nh->nh_oif != onh->nh_oif ||
181 nh->nh_gw != onh->nh_gw ||
182 nh->nh_scope != onh->nh_scope ||
183#ifdef CONFIG_IP_ROUTE_MULTIPATH
184 nh->nh_weight != onh->nh_weight ||
185#endif
186#ifdef CONFIG_NET_CLS_ROUTE
187 nh->nh_tclassid != onh->nh_tclassid ||
188#endif
189 ((nh->nh_flags^onh->nh_flags)&~RTNH_F_DEAD))
190 return -1;
191 onh++;
192 } endfor_nexthops(fi);
193 return 0;
194}
195
196static inline unsigned int fib_devindex_hashfn(unsigned int val)
197{
198 unsigned int mask = DEVINDEX_HASHSIZE - 1;
199
200 return (val ^
201 (val >> DEVINDEX_HASHBITS) ^
202 (val >> (DEVINDEX_HASHBITS * 2))) & mask;
203}
204
205static inline unsigned int fib_info_hashfn(const struct fib_info *fi)
206{
207 unsigned int mask = (fib_hash_size - 1);
208 unsigned int val = fi->fib_nhs;
209
210 val ^= fi->fib_protocol;
211 val ^= (__force u32)fi->fib_prefsrc;
212 val ^= fi->fib_priority;
213 for_nexthops(fi) {
214 val ^= fib_devindex_hashfn(nh->nh_oif);
215 } endfor_nexthops(fi)
216
217 return (val ^ (val >> 7) ^ (val >> 12)) & mask;
218}
219
220static struct fib_info *fib_find_info(const struct fib_info *nfi)
221{
222 struct hlist_head *head;
223 struct hlist_node *node;
224 struct fib_info *fi;
225 unsigned int hash;
226
227 hash = fib_info_hashfn(nfi);
228 head = &fib_info_hash[hash];
229
230 hlist_for_each_entry(fi, node, head, fib_hash) {
231 if (fi->fib_net != nfi->fib_net)
232 continue;
233 if (fi->fib_nhs != nfi->fib_nhs)
234 continue;
235 if (nfi->fib_protocol == fi->fib_protocol &&
236 nfi->fib_prefsrc == fi->fib_prefsrc &&
237 nfi->fib_priority == fi->fib_priority &&
238 memcmp(nfi->fib_metrics, fi->fib_metrics,
239 sizeof(fi->fib_metrics)) == 0 &&
240 ((nfi->fib_flags^fi->fib_flags)&~RTNH_F_DEAD) == 0 &&
241 (nfi->fib_nhs == 0 || nh_comp(fi, nfi) == 0))
242 return fi;
243 }
244
245 return NULL;
246}
247
248
249
250
251
252int ip_fib_check_default(__be32 gw, struct net_device *dev)
253{
254 struct hlist_head *head;
255 struct hlist_node *node;
256 struct fib_nh *nh;
257 unsigned int hash;
258
259 spin_lock(&fib_info_lock);
260
261 hash = fib_devindex_hashfn(dev->ifindex);
262 head = &fib_info_devhash[hash];
263 hlist_for_each_entry(nh, node, head, nh_hash) {
264 if (nh->nh_dev == dev &&
265 nh->nh_gw == gw &&
266 !(nh->nh_flags&RTNH_F_DEAD)) {
267 spin_unlock(&fib_info_lock);
268 return 0;
269 }
270 }
271
272 spin_unlock(&fib_info_lock);
273
274 return -1;
275}
276
277static inline size_t fib_nlmsg_size(struct fib_info *fi)
278{
279 size_t payload = NLMSG_ALIGN(sizeof(struct rtmsg))
280 + nla_total_size(4)
281 + nla_total_size(4)
282 + nla_total_size(4)
283 + nla_total_size(4);
284
285
286 payload += nla_total_size((RTAX_MAX * nla_total_size(4)));
287
288 if (fi->fib_nhs) {
289
290
291
292 size_t nhsize = nla_total_size(sizeof(struct rtnexthop));
293
294
295 nhsize += 2 * nla_total_size(4);
296
297
298 payload += nla_total_size(fi->fib_nhs * nhsize);
299 }
300
301 return payload;
302}
303
304void rtmsg_fib(int event, __be32 key, struct fib_alias *fa,
305 int dst_len, u32 tb_id, struct nl_info *info,
306 unsigned int nlm_flags)
307{
308 struct sk_buff *skb;
309 u32 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
310 int err = -ENOBUFS;
311
312 skb = nlmsg_new(fib_nlmsg_size(fa->fa_info), GFP_KERNEL);
313 if (skb == NULL)
314 goto errout;
315
316 err = fib_dump_info(skb, info->pid, seq, event, tb_id,
317 fa->fa_type, fa->fa_scope, key, dst_len,
318 fa->fa_tos, fa->fa_info, nlm_flags);
319 if (err < 0) {
320
321 WARN_ON(err == -EMSGSIZE);
322 kfree_skb(skb);
323 goto errout;
324 }
325 rtnl_notify(skb, info->nl_net, info->pid, RTNLGRP_IPV4_ROUTE,
326 info->nlh, GFP_KERNEL);
327 return;
328errout:
329 if (err < 0)
330 rtnl_set_sk_err(info->nl_net, RTNLGRP_IPV4_ROUTE, err);
331}
332
333
334
335
336struct fib_alias *fib_find_alias(struct list_head *fah, u8 tos, u32 prio)
337{
338 if (fah) {
339 struct fib_alias *fa;
340 list_for_each_entry(fa, fah, fa_list) {
341 if (fa->fa_tos > tos)
342 continue;
343 if (fa->fa_info->fib_priority >= prio ||
344 fa->fa_tos < tos)
345 return fa;
346 }
347 }
348 return NULL;
349}
350
351int fib_detect_death(struct fib_info *fi, int order,
352 struct fib_info **last_resort, int *last_idx, int dflt)
353{
354 struct neighbour *n;
355 int state = NUD_NONE;
356
357 n = neigh_lookup(&arp_tbl, &fi->fib_nh[0].nh_gw, fi->fib_dev);
358 if (n) {
359 state = n->nud_state;
360 neigh_release(n);
361 }
362 if (state == NUD_REACHABLE)
363 return 0;
364 if ((state&NUD_VALID) && order != dflt)
365 return 0;
366 if ((state&NUD_VALID) ||
367 (*last_idx<0 && order > dflt)) {
368 *last_resort = fi;
369 *last_idx = order;
370 }
371 return 1;
372}
373
374#ifdef CONFIG_IP_ROUTE_MULTIPATH
375
376static int fib_count_nexthops(struct rtnexthop *rtnh, int remaining)
377{
378 int nhs = 0;
379
380 while (rtnh_ok(rtnh, remaining)) {
381 nhs++;
382 rtnh = rtnh_next(rtnh, &remaining);
383 }
384
385
386 return remaining > 0 ? 0 : nhs;
387}
388
389static int fib_get_nhs(struct fib_info *fi, struct rtnexthop *rtnh,
390 int remaining, struct fib_config *cfg)
391{
392 change_nexthops(fi) {
393 int attrlen;
394
395 if (!rtnh_ok(rtnh, remaining))
396 return -EINVAL;
397
398 nh->nh_flags = (cfg->fc_flags & ~0xFF) | rtnh->rtnh_flags;
399 nh->nh_oif = rtnh->rtnh_ifindex;
400 nh->nh_weight = rtnh->rtnh_hops + 1;
401
402 attrlen = rtnh_attrlen(rtnh);
403 if (attrlen > 0) {
404 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
405
406 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
407 nh->nh_gw = nla ? nla_get_be32(nla) : 0;
408#ifdef CONFIG_NET_CLS_ROUTE
409 nla = nla_find(attrs, attrlen, RTA_FLOW);
410 nh->nh_tclassid = nla ? nla_get_u32(nla) : 0;
411#endif
412 }
413
414 rtnh = rtnh_next(rtnh, &remaining);
415 } endfor_nexthops(fi);
416
417 return 0;
418}
419
420#endif
421
422int fib_nh_match(struct fib_config *cfg, struct fib_info *fi)
423{
424#ifdef CONFIG_IP_ROUTE_MULTIPATH
425 struct rtnexthop *rtnh;
426 int remaining;
427#endif
428
429 if (cfg->fc_priority && cfg->fc_priority != fi->fib_priority)
430 return 1;
431
432 if (cfg->fc_oif || cfg->fc_gw) {
433 if ((!cfg->fc_oif || cfg->fc_oif == fi->fib_nh->nh_oif) &&
434 (!cfg->fc_gw || cfg->fc_gw == fi->fib_nh->nh_gw))
435 return 0;
436 return 1;
437 }
438
439#ifdef CONFIG_IP_ROUTE_MULTIPATH
440 if (cfg->fc_mp == NULL)
441 return 0;
442
443 rtnh = cfg->fc_mp;
444 remaining = cfg->fc_mp_len;
445
446 for_nexthops(fi) {
447 int attrlen;
448
449 if (!rtnh_ok(rtnh, remaining))
450 return -EINVAL;
451
452 if (rtnh->rtnh_ifindex && rtnh->rtnh_ifindex != nh->nh_oif)
453 return 1;
454
455 attrlen = rtnh_attrlen(rtnh);
456 if (attrlen < 0) {
457 struct nlattr *nla, *attrs = rtnh_attrs(rtnh);
458
459 nla = nla_find(attrs, attrlen, RTA_GATEWAY);
460 if (nla && nla_get_be32(nla) != nh->nh_gw)
461 return 1;
462#ifdef CONFIG_NET_CLS_ROUTE
463 nla = nla_find(attrs, attrlen, RTA_FLOW);
464 if (nla && nla_get_u32(nla) != nh->nh_tclassid)
465 return 1;
466#endif
467 }
468
469 rtnh = rtnh_next(rtnh, &remaining);
470 } endfor_nexthops(fi);
471#endif
472 return 0;
473}
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520static int fib_check_nh(struct fib_config *cfg, struct fib_info *fi,
521 struct fib_nh *nh)
522{
523 int err;
524 struct net *net;
525
526 net = cfg->fc_nlinfo.nl_net;
527 if (nh->nh_gw) {
528 struct fib_result res;
529
530#ifdef CONFIG_IP_ROUTE_PERVASIVE
531 if (nh->nh_flags&RTNH_F_PERVASIVE)
532 return 0;
533#endif
534 if (nh->nh_flags&RTNH_F_ONLINK) {
535 struct net_device *dev;
536
537 if (cfg->fc_scope >= RT_SCOPE_LINK)
538 return -EINVAL;
539 if (inet_addr_type(net, nh->nh_gw) != RTN_UNICAST)
540 return -EINVAL;
541 if ((dev = __dev_get_by_index(net, nh->nh_oif)) == NULL)
542 return -ENODEV;
543 if (!(dev->flags&IFF_UP))
544 return -ENETDOWN;
545 nh->nh_dev = dev;
546 dev_hold(dev);
547 nh->nh_scope = RT_SCOPE_LINK;
548 return 0;
549 }
550 {
551 struct flowi fl = {
552 .nl_u = {
553 .ip4_u = {
554 .daddr = nh->nh_gw,
555 .scope = cfg->fc_scope + 1,
556 },
557 },
558 .oif = nh->nh_oif,
559 };
560
561
562 if (fl.fl4_scope < RT_SCOPE_LINK)
563 fl.fl4_scope = RT_SCOPE_LINK;
564 if ((err = fib_lookup(net, &fl, &res)) != 0)
565 return err;
566 }
567 err = -EINVAL;
568 if (res.type != RTN_UNICAST && res.type != RTN_LOCAL)
569 goto out;
570 nh->nh_scope = res.scope;
571 nh->nh_oif = FIB_RES_OIF(res);
572 if ((nh->nh_dev = FIB_RES_DEV(res)) == NULL)
573 goto out;
574 dev_hold(nh->nh_dev);
575 err = -ENETDOWN;
576 if (!(nh->nh_dev->flags & IFF_UP))
577 goto out;
578 err = 0;
579out:
580 fib_res_put(&res);
581 return err;
582 } else {
583 struct in_device *in_dev;
584
585 if (nh->nh_flags&(RTNH_F_PERVASIVE|RTNH_F_ONLINK))
586 return -EINVAL;
587
588 in_dev = inetdev_by_index(net, nh->nh_oif);
589 if (in_dev == NULL)
590 return -ENODEV;
591 if (!(in_dev->dev->flags&IFF_UP)) {
592 in_dev_put(in_dev);
593 return -ENETDOWN;
594 }
595 nh->nh_dev = in_dev->dev;
596 dev_hold(nh->nh_dev);
597 nh->nh_scope = RT_SCOPE_HOST;
598 in_dev_put(in_dev);
599 }
600 return 0;
601}
602
603static inline unsigned int fib_laddr_hashfn(__be32 val)
604{
605 unsigned int mask = (fib_hash_size - 1);
606
607 return ((__force u32)val ^ ((__force u32)val >> 7) ^ ((__force u32)val >> 14)) & mask;
608}
609
610static struct hlist_head *fib_hash_alloc(int bytes)
611{
612 if (bytes <= PAGE_SIZE)
613 return kzalloc(bytes, GFP_KERNEL);
614 else
615 return (struct hlist_head *)
616 __get_free_pages(GFP_KERNEL | __GFP_ZERO, get_order(bytes));
617}
618
619static void fib_hash_free(struct hlist_head *hash, int bytes)
620{
621 if (!hash)
622 return;
623
624 if (bytes <= PAGE_SIZE)
625 kfree(hash);
626 else
627 free_pages((unsigned long) hash, get_order(bytes));
628}
629
630static void fib_hash_move(struct hlist_head *new_info_hash,
631 struct hlist_head *new_laddrhash,
632 unsigned int new_size)
633{
634 struct hlist_head *old_info_hash, *old_laddrhash;
635 unsigned int old_size = fib_hash_size;
636 unsigned int i, bytes;
637
638 spin_lock_bh(&fib_info_lock);
639 old_info_hash = fib_info_hash;
640 old_laddrhash = fib_info_laddrhash;
641 fib_hash_size = new_size;
642
643 for (i = 0; i < old_size; i++) {
644 struct hlist_head *head = &fib_info_hash[i];
645 struct hlist_node *node, *n;
646 struct fib_info *fi;
647
648 hlist_for_each_entry_safe(fi, node, n, head, fib_hash) {
649 struct hlist_head *dest;
650 unsigned int new_hash;
651
652 hlist_del(&fi->fib_hash);
653
654 new_hash = fib_info_hashfn(fi);
655 dest = &new_info_hash[new_hash];
656 hlist_add_head(&fi->fib_hash, dest);
657 }
658 }
659 fib_info_hash = new_info_hash;
660
661 for (i = 0; i < old_size; i++) {
662 struct hlist_head *lhead = &fib_info_laddrhash[i];
663 struct hlist_node *node, *n;
664 struct fib_info *fi;
665
666 hlist_for_each_entry_safe(fi, node, n, lhead, fib_lhash) {
667 struct hlist_head *ldest;
668 unsigned int new_hash;
669
670 hlist_del(&fi->fib_lhash);
671
672 new_hash = fib_laddr_hashfn(fi->fib_prefsrc);
673 ldest = &new_laddrhash[new_hash];
674 hlist_add_head(&fi->fib_lhash, ldest);
675 }
676 }
677 fib_info_laddrhash = new_laddrhash;
678
679 spin_unlock_bh(&fib_info_lock);
680
681 bytes = old_size * sizeof(struct hlist_head *);
682 fib_hash_free(old_info_hash, bytes);
683 fib_hash_free(old_laddrhash, bytes);
684}
685
686struct fib_info *fib_create_info(struct fib_config *cfg)
687{
688 int err;
689 struct fib_info *fi = NULL;
690 struct fib_info *ofi;
691 int nhs = 1;
692 struct net *net = cfg->fc_nlinfo.nl_net;
693
694
695 if (fib_props[cfg->fc_type].scope > cfg->fc_scope)
696 goto err_inval;
697
698#ifdef CONFIG_IP_ROUTE_MULTIPATH
699 if (cfg->fc_mp) {
700 nhs = fib_count_nexthops(cfg->fc_mp, cfg->fc_mp_len);
701 if (nhs == 0)
702 goto err_inval;
703 }
704#endif
705
706 err = -ENOBUFS;
707 if (fib_info_cnt >= fib_hash_size) {
708 unsigned int new_size = fib_hash_size << 1;
709 struct hlist_head *new_info_hash;
710 struct hlist_head *new_laddrhash;
711 unsigned int bytes;
712
713 if (!new_size)
714 new_size = 1;
715 bytes = new_size * sizeof(struct hlist_head *);
716 new_info_hash = fib_hash_alloc(bytes);
717 new_laddrhash = fib_hash_alloc(bytes);
718 if (!new_info_hash || !new_laddrhash) {
719 fib_hash_free(new_info_hash, bytes);
720 fib_hash_free(new_laddrhash, bytes);
721 } else
722 fib_hash_move(new_info_hash, new_laddrhash, new_size);
723
724 if (!fib_hash_size)
725 goto failure;
726 }
727
728 fi = kzalloc(sizeof(*fi)+nhs*sizeof(struct fib_nh), GFP_KERNEL);
729 if (fi == NULL)
730 goto failure;
731 fib_info_cnt++;
732
733 fi->fib_net = hold_net(net);
734 fi->fib_protocol = cfg->fc_protocol;
735 fi->fib_flags = cfg->fc_flags;
736 fi->fib_priority = cfg->fc_priority;
737 fi->fib_prefsrc = cfg->fc_prefsrc;
738
739 fi->fib_nhs = nhs;
740 change_nexthops(fi) {
741 nh->nh_parent = fi;
742 } endfor_nexthops(fi)
743
744 if (cfg->fc_mx) {
745 struct nlattr *nla;
746 int remaining;
747
748 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
749 int type = nla_type(nla);
750
751 if (type) {
752 if (type > RTAX_MAX)
753 goto err_inval;
754 fi->fib_metrics[type - 1] = nla_get_u32(nla);
755 }
756 }
757 }
758
759 if (cfg->fc_mp) {
760#ifdef CONFIG_IP_ROUTE_MULTIPATH
761 err = fib_get_nhs(fi, cfg->fc_mp, cfg->fc_mp_len, cfg);
762 if (err != 0)
763 goto failure;
764 if (cfg->fc_oif && fi->fib_nh->nh_oif != cfg->fc_oif)
765 goto err_inval;
766 if (cfg->fc_gw && fi->fib_nh->nh_gw != cfg->fc_gw)
767 goto err_inval;
768#ifdef CONFIG_NET_CLS_ROUTE
769 if (cfg->fc_flow && fi->fib_nh->nh_tclassid != cfg->fc_flow)
770 goto err_inval;
771#endif
772#else
773 goto err_inval;
774#endif
775 } else {
776 struct fib_nh *nh = fi->fib_nh;
777
778 nh->nh_oif = cfg->fc_oif;
779 nh->nh_gw = cfg->fc_gw;
780 nh->nh_flags = cfg->fc_flags;
781#ifdef CONFIG_NET_CLS_ROUTE
782 nh->nh_tclassid = cfg->fc_flow;
783#endif
784#ifdef CONFIG_IP_ROUTE_MULTIPATH
785 nh->nh_weight = 1;
786#endif
787 }
788
789 if (fib_props[cfg->fc_type].error) {
790 if (cfg->fc_gw || cfg->fc_oif || cfg->fc_mp)
791 goto err_inval;
792 goto link_it;
793 }
794
795 if (cfg->fc_scope > RT_SCOPE_HOST)
796 goto err_inval;
797
798 if (cfg->fc_scope == RT_SCOPE_HOST) {
799 struct fib_nh *nh = fi->fib_nh;
800
801
802 if (nhs != 1 || nh->nh_gw)
803 goto err_inval;
804 nh->nh_scope = RT_SCOPE_NOWHERE;
805 nh->nh_dev = dev_get_by_index(net, fi->fib_nh->nh_oif);
806 err = -ENODEV;
807 if (nh->nh_dev == NULL)
808 goto failure;
809 } else {
810 change_nexthops(fi) {
811 if ((err = fib_check_nh(cfg, fi, nh)) != 0)
812 goto failure;
813 } endfor_nexthops(fi)
814 }
815
816 if (fi->fib_prefsrc) {
817 if (cfg->fc_type != RTN_LOCAL || !cfg->fc_dst ||
818 fi->fib_prefsrc != cfg->fc_dst)
819 if (inet_addr_type(net, fi->fib_prefsrc) != RTN_LOCAL)
820 goto err_inval;
821 }
822
823link_it:
824 if ((ofi = fib_find_info(fi)) != NULL) {
825 fi->fib_dead = 1;
826 free_fib_info(fi);
827 ofi->fib_treeref++;
828 return ofi;
829 }
830
831 fi->fib_treeref++;
832 atomic_inc(&fi->fib_clntref);
833 spin_lock_bh(&fib_info_lock);
834 hlist_add_head(&fi->fib_hash,
835 &fib_info_hash[fib_info_hashfn(fi)]);
836 if (fi->fib_prefsrc) {
837 struct hlist_head *head;
838
839 head = &fib_info_laddrhash[fib_laddr_hashfn(fi->fib_prefsrc)];
840 hlist_add_head(&fi->fib_lhash, head);
841 }
842 change_nexthops(fi) {
843 struct hlist_head *head;
844 unsigned int hash;
845
846 if (!nh->nh_dev)
847 continue;
848 hash = fib_devindex_hashfn(nh->nh_dev->ifindex);
849 head = &fib_info_devhash[hash];
850 hlist_add_head(&nh->nh_hash, head);
851 } endfor_nexthops(fi)
852 spin_unlock_bh(&fib_info_lock);
853 return fi;
854
855err_inval:
856 err = -EINVAL;
857
858failure:
859 if (fi) {
860 fi->fib_dead = 1;
861 free_fib_info(fi);
862 }
863
864 return ERR_PTR(err);
865}
866
867
868int fib_semantic_match(struct list_head *head, const struct flowi *flp,
869 struct fib_result *res, int prefixlen)
870{
871 struct fib_alias *fa;
872 int nh_sel = 0;
873
874 list_for_each_entry_rcu(fa, head, fa_list) {
875 int err;
876
877 if (fa->fa_tos &&
878 fa->fa_tos != flp->fl4_tos)
879 continue;
880
881 if (fa->fa_scope < flp->fl4_scope)
882 continue;
883
884 fa->fa_state |= FA_S_ACCESSED;
885
886 err = fib_props[fa->fa_type].error;
887 if (err == 0) {
888 struct fib_info *fi = fa->fa_info;
889
890 if (fi->fib_flags & RTNH_F_DEAD)
891 continue;
892
893 switch (fa->fa_type) {
894 case RTN_UNICAST:
895 case RTN_LOCAL:
896 case RTN_BROADCAST:
897 case RTN_ANYCAST:
898 case RTN_MULTICAST:
899 for_nexthops(fi) {
900 if (nh->nh_flags&RTNH_F_DEAD)
901 continue;
902 if (!flp->oif || flp->oif == nh->nh_oif)
903 break;
904 }
905#ifdef CONFIG_IP_ROUTE_MULTIPATH
906 if (nhsel < fi->fib_nhs) {
907 nh_sel = nhsel;
908 goto out_fill_res;
909 }
910#else
911 if (nhsel < 1) {
912 goto out_fill_res;
913 }
914#endif
915 endfor_nexthops(fi);
916 continue;
917
918 default:
919 printk(KERN_WARNING "fib_semantic_match bad type %#x\n",
920 fa->fa_type);
921 return -EINVAL;
922 }
923 }
924 return err;
925 }
926 return 1;
927
928out_fill_res:
929 res->prefixlen = prefixlen;
930 res->nh_sel = nh_sel;
931 res->type = fa->fa_type;
932 res->scope = fa->fa_scope;
933 res->fi = fa->fa_info;
934 atomic_inc(&res->fi->fib_clntref);
935 return 0;
936}
937
938
939
940__be32 __fib_res_prefsrc(struct fib_result *res)
941{
942 return inet_select_addr(FIB_RES_DEV(*res), FIB_RES_GW(*res), res->scope);
943}
944
945int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event,
946 u32 tb_id, u8 type, u8 scope, __be32 dst, int dst_len, u8 tos,
947 struct fib_info *fi, unsigned int flags)
948{
949 struct nlmsghdr *nlh;
950 struct rtmsg *rtm;
951
952 nlh = nlmsg_put(skb, pid, seq, event, sizeof(*rtm), flags);
953 if (nlh == NULL)
954 return -EMSGSIZE;
955
956 rtm = nlmsg_data(nlh);
957 rtm->rtm_family = AF_INET;
958 rtm->rtm_dst_len = dst_len;
959 rtm->rtm_src_len = 0;
960 rtm->rtm_tos = tos;
961 if (tb_id < 256)
962 rtm->rtm_table = tb_id;
963 else
964 rtm->rtm_table = RT_TABLE_COMPAT;
965 NLA_PUT_U32(skb, RTA_TABLE, tb_id);
966 rtm->rtm_type = type;
967 rtm->rtm_flags = fi->fib_flags;
968 rtm->rtm_scope = scope;
969 rtm->rtm_protocol = fi->fib_protocol;
970
971 if (rtm->rtm_dst_len)
972 NLA_PUT_BE32(skb, RTA_DST, dst);
973
974 if (fi->fib_priority)
975 NLA_PUT_U32(skb, RTA_PRIORITY, fi->fib_priority);
976
977 if (rtnetlink_put_metrics(skb, fi->fib_metrics) < 0)
978 goto nla_put_failure;
979
980 if (fi->fib_prefsrc)
981 NLA_PUT_BE32(skb, RTA_PREFSRC, fi->fib_prefsrc);
982
983 if (fi->fib_nhs == 1) {
984 if (fi->fib_nh->nh_gw)
985 NLA_PUT_BE32(skb, RTA_GATEWAY, fi->fib_nh->nh_gw);
986
987 if (fi->fib_nh->nh_oif)
988 NLA_PUT_U32(skb, RTA_OIF, fi->fib_nh->nh_oif);
989#ifdef CONFIG_NET_CLS_ROUTE
990 if (fi->fib_nh[0].nh_tclassid)
991 NLA_PUT_U32(skb, RTA_FLOW, fi->fib_nh[0].nh_tclassid);
992#endif
993 }
994#ifdef CONFIG_IP_ROUTE_MULTIPATH
995 if (fi->fib_nhs > 1) {
996 struct rtnexthop *rtnh;
997 struct nlattr *mp;
998
999 mp = nla_nest_start(skb, RTA_MULTIPATH);
1000 if (mp == NULL)
1001 goto nla_put_failure;
1002
1003 for_nexthops(fi) {
1004 rtnh = nla_reserve_nohdr(skb, sizeof(*rtnh));
1005 if (rtnh == NULL)
1006 goto nla_put_failure;
1007
1008 rtnh->rtnh_flags = nh->nh_flags & 0xFF;
1009 rtnh->rtnh_hops = nh->nh_weight - 1;
1010 rtnh->rtnh_ifindex = nh->nh_oif;
1011
1012 if (nh->nh_gw)
1013 NLA_PUT_BE32(skb, RTA_GATEWAY, nh->nh_gw);
1014#ifdef CONFIG_NET_CLS_ROUTE
1015 if (nh->nh_tclassid)
1016 NLA_PUT_U32(skb, RTA_FLOW, nh->nh_tclassid);
1017#endif
1018
1019 rtnh->rtnh_len = nlmsg_get_pos(skb) - (void *) rtnh;
1020 } endfor_nexthops(fi);
1021
1022 nla_nest_end(skb, mp);
1023 }
1024#endif
1025 return nlmsg_end(skb, nlh);
1026
1027nla_put_failure:
1028 nlmsg_cancel(skb, nlh);
1029 return -EMSGSIZE;
1030}
1031
1032
1033
1034
1035
1036
1037
1038int fib_sync_down_addr(struct net *net, __be32 local)
1039{
1040 int ret = 0;
1041 unsigned int hash = fib_laddr_hashfn(local);
1042 struct hlist_head *head = &fib_info_laddrhash[hash];
1043 struct hlist_node *node;
1044 struct fib_info *fi;
1045
1046 if (fib_info_laddrhash == NULL || local == 0)
1047 return 0;
1048
1049 hlist_for_each_entry(fi, node, head, fib_lhash) {
1050 if (fi->fib_net != net)
1051 continue;
1052 if (fi->fib_prefsrc == local) {
1053 fi->fib_flags |= RTNH_F_DEAD;
1054 ret++;
1055 }
1056 }
1057 return ret;
1058}
1059
1060int fib_sync_down_dev(struct net_device *dev, int force)
1061{
1062 int ret = 0;
1063 int scope = RT_SCOPE_NOWHERE;
1064 struct fib_info *prev_fi = NULL;
1065 unsigned int hash = fib_devindex_hashfn(dev->ifindex);
1066 struct hlist_head *head = &fib_info_devhash[hash];
1067 struct hlist_node *node;
1068 struct fib_nh *nh;
1069
1070 if (force)
1071 scope = -1;
1072
1073 hlist_for_each_entry(nh, node, head, nh_hash) {
1074 struct fib_info *fi = nh->nh_parent;
1075 int dead;
1076
1077 BUG_ON(!fi->fib_nhs);
1078 if (nh->nh_dev != dev || fi == prev_fi)
1079 continue;
1080 prev_fi = fi;
1081 dead = 0;
1082 change_nexthops(fi) {
1083 if (nh->nh_flags&RTNH_F_DEAD)
1084 dead++;
1085 else if (nh->nh_dev == dev &&
1086 nh->nh_scope != scope) {
1087 nh->nh_flags |= RTNH_F_DEAD;
1088#ifdef CONFIG_IP_ROUTE_MULTIPATH
1089 spin_lock_bh(&fib_multipath_lock);
1090 fi->fib_power -= nh->nh_power;
1091 nh->nh_power = 0;
1092 spin_unlock_bh(&fib_multipath_lock);
1093#endif
1094 dead++;
1095 }
1096#ifdef CONFIG_IP_ROUTE_MULTIPATH
1097 if (force > 1 && nh->nh_dev == dev) {
1098 dead = fi->fib_nhs;
1099 break;
1100 }
1101#endif
1102 } endfor_nexthops(fi)
1103 if (dead == fi->fib_nhs) {
1104 fi->fib_flags |= RTNH_F_DEAD;
1105 ret++;
1106 }
1107 }
1108
1109 return ret;
1110}
1111
1112#ifdef CONFIG_IP_ROUTE_MULTIPATH
1113
1114
1115
1116
1117
1118
1119int fib_sync_up(struct net_device *dev)
1120{
1121 struct fib_info *prev_fi;
1122 unsigned int hash;
1123 struct hlist_head *head;
1124 struct hlist_node *node;
1125 struct fib_nh *nh;
1126 int ret;
1127
1128 if (!(dev->flags&IFF_UP))
1129 return 0;
1130
1131 prev_fi = NULL;
1132 hash = fib_devindex_hashfn(dev->ifindex);
1133 head = &fib_info_devhash[hash];
1134 ret = 0;
1135
1136 hlist_for_each_entry(nh, node, head, nh_hash) {
1137 struct fib_info *fi = nh->nh_parent;
1138 int alive;
1139
1140 BUG_ON(!fi->fib_nhs);
1141 if (nh->nh_dev != dev || fi == prev_fi)
1142 continue;
1143
1144 prev_fi = fi;
1145 alive = 0;
1146 change_nexthops(fi) {
1147 if (!(nh->nh_flags&RTNH_F_DEAD)) {
1148 alive++;
1149 continue;
1150 }
1151 if (nh->nh_dev == NULL || !(nh->nh_dev->flags&IFF_UP))
1152 continue;
1153 if (nh->nh_dev != dev || !__in_dev_get_rtnl(dev))
1154 continue;
1155 alive++;
1156 spin_lock_bh(&fib_multipath_lock);
1157 nh->nh_power = 0;
1158 nh->nh_flags &= ~RTNH_F_DEAD;
1159 spin_unlock_bh(&fib_multipath_lock);
1160 } endfor_nexthops(fi)
1161
1162 if (alive > 0) {
1163 fi->fib_flags &= ~RTNH_F_DEAD;
1164 ret++;
1165 }
1166 }
1167
1168 return ret;
1169}
1170
1171
1172
1173
1174
1175
1176void fib_select_multipath(const struct flowi *flp, struct fib_result *res)
1177{
1178 struct fib_info *fi = res->fi;
1179 int w;
1180
1181 spin_lock_bh(&fib_multipath_lock);
1182 if (fi->fib_power <= 0) {
1183 int power = 0;
1184 change_nexthops(fi) {
1185 if (!(nh->nh_flags&RTNH_F_DEAD)) {
1186 power += nh->nh_weight;
1187 nh->nh_power = nh->nh_weight;
1188 }
1189 } endfor_nexthops(fi);
1190 fi->fib_power = power;
1191 if (power <= 0) {
1192 spin_unlock_bh(&fib_multipath_lock);
1193
1194 res->nh_sel = 0;
1195 return;
1196 }
1197 }
1198
1199
1200
1201
1202
1203
1204 w = jiffies % fi->fib_power;
1205
1206 change_nexthops(fi) {
1207 if (!(nh->nh_flags&RTNH_F_DEAD) && nh->nh_power) {
1208 if ((w -= nh->nh_power) <= 0) {
1209 nh->nh_power--;
1210 fi->fib_power--;
1211 res->nh_sel = nhsel;
1212 spin_unlock_bh(&fib_multipath_lock);
1213 return;
1214 }
1215 }
1216 } endfor_nexthops(fi);
1217
1218
1219 res->nh_sel = 0;
1220 spin_unlock_bh(&fib_multipath_lock);
1221}
1222#endif
1223
