aboutsummaryrefslogtreecommitdiffstats
path: root/net/ipv6/route.c
diff options
context:
space:
mode:
Diffstat (limited to 'net/ipv6/route.c')
-rw-r--r--net/ipv6/route.c1071
1 files changed, 621 insertions, 450 deletions
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 8a777932786d..d6b4b4f48d18 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -22,10 +22,11 @@
22 * routers in REACHABLE, STALE, DELAY or PROBE states). 22 * routers in REACHABLE, STALE, DELAY or PROBE states).
23 * - always select the same router if it is (probably) 23 * - always select the same router if it is (probably)
24 * reachable. otherwise, round-robin the list. 24 * reachable. otherwise, round-robin the list.
25 * Ville Nuorvala
26 * Fixed routing subtrees.
25 */ 27 */
26 28
27#include <linux/capability.h> 29#include <linux/capability.h>
28#include <linux/config.h>
29#include <linux/errno.h> 30#include <linux/errno.h>
30#include <linux/types.h> 31#include <linux/types.h>
31#include <linux/times.h> 32#include <linux/times.h>
@@ -36,7 +37,6 @@
36#include <linux/netdevice.h> 37#include <linux/netdevice.h>
37#include <linux/in6.h> 38#include <linux/in6.h>
38#include <linux/init.h> 39#include <linux/init.h>
39#include <linux/netlink.h>
40#include <linux/if_arp.h> 40#include <linux/if_arp.h>
41 41
42#ifdef CONFIG_PROC_FS 42#ifdef CONFIG_PROC_FS
@@ -54,6 +54,8 @@
54#include <linux/rtnetlink.h> 54#include <linux/rtnetlink.h>
55#include <net/dst.h> 55#include <net/dst.h>
56#include <net/xfrm.h> 56#include <net/xfrm.h>
57#include <net/netevent.h>
58#include <net/netlink.h>
57 59
58#include <asm/uaccess.h> 60#include <asm/uaccess.h>
59 61
@@ -74,9 +76,6 @@
74 76
75#define CLONE_OFFLINK_ROUTE 0 77#define CLONE_OFFLINK_ROUTE 0
76 78
77#define RT6_SELECT_F_IFACE 0x1
78#define RT6_SELECT_F_REACHABLE 0x2
79
80static int ip6_rt_max_size = 4096; 79static int ip6_rt_max_size = 4096;
81static int ip6_rt_gc_min_interval = HZ / 2; 80static int ip6_rt_gc_min_interval = HZ / 2;
82static int ip6_rt_gc_timeout = 60*HZ; 81static int ip6_rt_gc_timeout = 60*HZ;
@@ -140,15 +139,49 @@ struct rt6_info ip6_null_entry = {
140 .rt6i_ref = ATOMIC_INIT(1), 139 .rt6i_ref = ATOMIC_INIT(1),
141}; 140};
142 141
143struct fib6_node ip6_routing_table = { 142#ifdef CONFIG_IPV6_MULTIPLE_TABLES
144 .leaf = &ip6_null_entry,
145 .fn_flags = RTN_ROOT | RTN_TL_ROOT | RTN_RTINFO,
146};
147 143
148/* Protects all the ip6 fib */ 144struct rt6_info ip6_prohibit_entry = {
145 .u = {
146 .dst = {
147 .__refcnt = ATOMIC_INIT(1),
148 .__use = 1,
149 .dev = &loopback_dev,
150 .obsolete = -1,
151 .error = -EACCES,
152 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
153 .input = ip6_pkt_discard,
154 .output = ip6_pkt_discard_out,
155 .ops = &ip6_dst_ops,
156 .path = (struct dst_entry*)&ip6_prohibit_entry,
157 }
158 },
159 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
160 .rt6i_metric = ~(u32) 0,
161 .rt6i_ref = ATOMIC_INIT(1),
162};
149 163
150DEFINE_RWLOCK(rt6_lock); 164struct rt6_info ip6_blk_hole_entry = {
165 .u = {
166 .dst = {
167 .__refcnt = ATOMIC_INIT(1),
168 .__use = 1,
169 .dev = &loopback_dev,
170 .obsolete = -1,
171 .error = -EINVAL,
172 .metrics = { [RTAX_HOPLIMIT - 1] = 255, },
173 .input = ip6_pkt_discard,
174 .output = ip6_pkt_discard_out,
175 .ops = &ip6_dst_ops,
176 .path = (struct dst_entry*)&ip6_blk_hole_entry,
177 }
178 },
179 .rt6i_flags = (RTF_REJECT | RTF_NONEXTHOP),
180 .rt6i_metric = ~(u32) 0,
181 .rt6i_ref = ATOMIC_INIT(1),
182};
151 183
184#endif
152 185
153/* allocate dst with ip6_dst_ops */ 186/* allocate dst with ip6_dst_ops */
154static __inline__ struct rt6_info *ip6_dst_alloc(void) 187static __inline__ struct rt6_info *ip6_dst_alloc(void)
@@ -188,8 +221,14 @@ static __inline__ int rt6_check_expired(const struct rt6_info *rt)
188 time_after(jiffies, rt->rt6i_expires)); 221 time_after(jiffies, rt->rt6i_expires));
189} 222}
190 223
224static inline int rt6_need_strict(struct in6_addr *daddr)
225{
226 return (ipv6_addr_type(daddr) &
227 (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL));
228}
229
191/* 230/*
192 * Route lookup. Any rt6_lock is implied. 231 * Route lookup. Any table->tb6_lock is implied.
193 */ 232 */
194 233
195static __inline__ struct rt6_info *rt6_device_match(struct rt6_info *rt, 234static __inline__ struct rt6_info *rt6_device_match(struct rt6_info *rt,
@@ -298,7 +337,7 @@ static int rt6_score_route(struct rt6_info *rt, int oif,
298 int m, n; 337 int m, n;
299 338
300 m = rt6_check_dev(rt, oif); 339 m = rt6_check_dev(rt, oif);
301 if (!m && (strict & RT6_SELECT_F_IFACE)) 340 if (!m && (strict & RT6_LOOKUP_F_IFACE))
302 return -1; 341 return -1;
303#ifdef CONFIG_IPV6_ROUTER_PREF 342#ifdef CONFIG_IPV6_ROUTER_PREF
304 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2; 343 m |= IPV6_DECODE_PREF(IPV6_EXTRACT_PREF(rt->rt6i_flags)) << 2;
@@ -306,7 +345,7 @@ static int rt6_score_route(struct rt6_info *rt, int oif,
306 n = rt6_check_neigh(rt); 345 n = rt6_check_neigh(rt);
307 if (n > 1) 346 if (n > 1)
308 m |= 16; 347 m |= 16;
309 else if (!n && strict & RT6_SELECT_F_REACHABLE) 348 else if (!n && strict & RT6_LOOKUP_F_REACHABLE)
310 return -1; 349 return -1;
311 return m; 350 return m;
312} 351}
@@ -346,10 +385,10 @@ static struct rt6_info *rt6_select(struct rt6_info **head, int oif,
346 } 385 }
347 386
348 if (!match && 387 if (!match &&
349 (strict & RT6_SELECT_F_REACHABLE) && 388 (strict & RT6_LOOKUP_F_REACHABLE) &&
350 last && last != rt0) { 389 last && last != rt0) {
351 /* no entries matched; do round-robin */ 390 /* no entries matched; do round-robin */
352 static spinlock_t lock = SPIN_LOCK_UNLOCKED; 391 static DEFINE_SPINLOCK(lock);
353 spin_lock(&lock); 392 spin_lock(&lock);
354 *head = rt0->u.next; 393 *head = rt0->u.next;
355 rt0->u.next = last->u.next; 394 rt0->u.next = last->u.next;
@@ -417,7 +456,7 @@ int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
417 rt = rt6_get_route_info(prefix, rinfo->prefix_len, gwaddr, dev->ifindex); 456 rt = rt6_get_route_info(prefix, rinfo->prefix_len, gwaddr, dev->ifindex);
418 457
419 if (rt && !lifetime) { 458 if (rt && !lifetime) {
420 ip6_del_rt(rt, NULL, NULL, NULL); 459 ip6_del_rt(rt);
421 rt = NULL; 460 rt = NULL;
422 } 461 }
423 462
@@ -441,44 +480,95 @@ int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
441} 480}
442#endif 481#endif
443 482
444struct rt6_info *rt6_lookup(struct in6_addr *daddr, struct in6_addr *saddr, 483#define BACKTRACK(saddr) \
445 int oif, int strict) 484do { \
485 if (rt == &ip6_null_entry) { \
486 struct fib6_node *pn; \
487 while (fn) { \
488 if (fn->fn_flags & RTN_TL_ROOT) \
489 goto out; \
490 pn = fn->parent; \
491 if (FIB6_SUBTREE(pn) && FIB6_SUBTREE(pn) != fn) \
492 fn = fib6_lookup(pn->subtree, NULL, saddr); \
493 else \
494 fn = pn; \
495 if (fn->fn_flags & RTN_RTINFO) \
496 goto restart; \
497 } \
498 } \
499} while(0)
500
501static struct rt6_info *ip6_pol_route_lookup(struct fib6_table *table,
502 struct flowi *fl, int flags)
446{ 503{
447 struct fib6_node *fn; 504 struct fib6_node *fn;
448 struct rt6_info *rt; 505 struct rt6_info *rt;
449 506
450 read_lock_bh(&rt6_lock); 507 read_lock_bh(&table->tb6_lock);
451 fn = fib6_lookup(&ip6_routing_table, daddr, saddr); 508 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
452 rt = rt6_device_match(fn->leaf, oif, strict); 509restart:
510 rt = fn->leaf;
511 rt = rt6_device_match(rt, fl->oif, flags);
512 BACKTRACK(&fl->fl6_src);
513out:
453 dst_hold(&rt->u.dst); 514 dst_hold(&rt->u.dst);
454 rt->u.dst.__use++; 515 read_unlock_bh(&table->tb6_lock);
455 read_unlock_bh(&rt6_lock);
456 516
457 rt->u.dst.lastuse = jiffies; 517 rt->u.dst.lastuse = jiffies;
458 if (rt->u.dst.error == 0) 518 rt->u.dst.__use++;
459 return rt; 519
460 dst_release(&rt->u.dst); 520 return rt;
521
522}
523
524struct rt6_info *rt6_lookup(struct in6_addr *daddr, struct in6_addr *saddr,
525 int oif, int strict)
526{
527 struct flowi fl = {
528 .oif = oif,
529 .nl_u = {
530 .ip6_u = {
531 .daddr = *daddr,
532 /* TODO: saddr */
533 },
534 },
535 };
536 struct dst_entry *dst;
537 int flags = strict ? RT6_LOOKUP_F_IFACE : 0;
538
539 dst = fib6_rule_lookup(&fl, flags, ip6_pol_route_lookup);
540 if (dst->error == 0)
541 return (struct rt6_info *) dst;
542
543 dst_release(dst);
544
461 return NULL; 545 return NULL;
462} 546}
463 547
464/* ip6_ins_rt is called with FREE rt6_lock. 548/* ip6_ins_rt is called with FREE table->tb6_lock.
465 It takes new route entry, the addition fails by any reason the 549 It takes new route entry, the addition fails by any reason the
466 route is freed. In any case, if caller does not hold it, it may 550 route is freed. In any case, if caller does not hold it, it may
467 be destroyed. 551 be destroyed.
468 */ 552 */
469 553
470int ip6_ins_rt(struct rt6_info *rt, struct nlmsghdr *nlh, 554static int __ip6_ins_rt(struct rt6_info *rt, struct nl_info *info)
471 void *_rtattr, struct netlink_skb_parms *req)
472{ 555{
473 int err; 556 int err;
557 struct fib6_table *table;
474 558
475 write_lock_bh(&rt6_lock); 559 table = rt->rt6i_table;
476 err = fib6_add(&ip6_routing_table, rt, nlh, _rtattr, req); 560 write_lock_bh(&table->tb6_lock);
477 write_unlock_bh(&rt6_lock); 561 err = fib6_add(&table->tb6_root, rt, info);
562 write_unlock_bh(&table->tb6_lock);
478 563
479 return err; 564 return err;
480} 565}
481 566
567int ip6_ins_rt(struct rt6_info *rt)
568{
569 return __ip6_ins_rt(rt, NULL);
570}
571
482static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, struct in6_addr *daddr, 572static struct rt6_info *rt6_alloc_cow(struct rt6_info *ort, struct in6_addr *daddr,
483 struct in6_addr *saddr) 573 struct in6_addr *saddr)
484{ 574{
@@ -532,51 +622,39 @@ static struct rt6_info *rt6_alloc_clone(struct rt6_info *ort, struct in6_addr *d
532 return rt; 622 return rt;
533} 623}
534 624
535#define BACKTRACK() \ 625static struct rt6_info *ip6_pol_route_input(struct fib6_table *table,
536if (rt == &ip6_null_entry) { \ 626 struct flowi *fl, int flags)
537 while ((fn = fn->parent) != NULL) { \
538 if (fn->fn_flags & RTN_ROOT) { \
539 goto out; \
540 } \
541 if (fn->fn_flags & RTN_RTINFO) \
542 goto restart; \
543 } \
544}
545
546
547void ip6_route_input(struct sk_buff *skb)
548{ 627{
549 struct fib6_node *fn; 628 struct fib6_node *fn;
550 struct rt6_info *rt, *nrt; 629 struct rt6_info *rt, *nrt;
551 int strict; 630 int strict = 0;
552 int attempts = 3; 631 int attempts = 3;
553 int err; 632 int err;
554 int reachable = RT6_SELECT_F_REACHABLE; 633 int reachable = RT6_LOOKUP_F_REACHABLE;
555 634
556 strict = ipv6_addr_type(&skb->nh.ipv6h->daddr) & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL) ? RT6_SELECT_F_IFACE : 0; 635 strict |= flags & RT6_LOOKUP_F_IFACE;
557 636
558relookup: 637relookup:
559 read_lock_bh(&rt6_lock); 638 read_lock_bh(&table->tb6_lock);
560 639
561restart_2: 640restart_2:
562 fn = fib6_lookup(&ip6_routing_table, &skb->nh.ipv6h->daddr, 641 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
563 &skb->nh.ipv6h->saddr);
564 642
565restart: 643restart:
566 rt = rt6_select(&fn->leaf, skb->dev->ifindex, strict | reachable); 644 rt = rt6_select(&fn->leaf, fl->iif, strict | reachable);
567 BACKTRACK(); 645 BACKTRACK(&fl->fl6_src);
568 if (rt == &ip6_null_entry || 646 if (rt == &ip6_null_entry ||
569 rt->rt6i_flags & RTF_CACHE) 647 rt->rt6i_flags & RTF_CACHE)
570 goto out; 648 goto out;
571 649
572 dst_hold(&rt->u.dst); 650 dst_hold(&rt->u.dst);
573 read_unlock_bh(&rt6_lock); 651 read_unlock_bh(&table->tb6_lock);
574 652
575 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP)) 653 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP))
576 nrt = rt6_alloc_cow(rt, &skb->nh.ipv6h->daddr, &skb->nh.ipv6h->saddr); 654 nrt = rt6_alloc_cow(rt, &fl->fl6_dst, &fl->fl6_src);
577 else { 655 else {
578#if CLONE_OFFLINK_ROUTE 656#if CLONE_OFFLINK_ROUTE
579 nrt = rt6_alloc_clone(rt, &skb->nh.ipv6h->daddr); 657 nrt = rt6_alloc_clone(rt, &fl->fl6_dst);
580#else 658#else
581 goto out2; 659 goto out2;
582#endif 660#endif
@@ -587,7 +665,7 @@ restart:
587 665
588 dst_hold(&rt->u.dst); 666 dst_hold(&rt->u.dst);
589 if (nrt) { 667 if (nrt) {
590 err = ip6_ins_rt(nrt, NULL, NULL, &NETLINK_CB(skb)); 668 err = ip6_ins_rt(nrt);
591 if (!err) 669 if (!err)
592 goto out2; 670 goto out2;
593 } 671 }
@@ -596,7 +674,7 @@ restart:
596 goto out2; 674 goto out2;
597 675
598 /* 676 /*
599 * Race condition! In the gap, when rt6_lock was 677 * Race condition! In the gap, when table->tb6_lock was
600 * released someone could insert this route. Relookup. 678 * released someone could insert this route. Relookup.
601 */ 679 */
602 dst_release(&rt->u.dst); 680 dst_release(&rt->u.dst);
@@ -608,40 +686,63 @@ out:
608 goto restart_2; 686 goto restart_2;
609 } 687 }
610 dst_hold(&rt->u.dst); 688 dst_hold(&rt->u.dst);
611 read_unlock_bh(&rt6_lock); 689 read_unlock_bh(&table->tb6_lock);
612out2: 690out2:
613 rt->u.dst.lastuse = jiffies; 691 rt->u.dst.lastuse = jiffies;
614 rt->u.dst.__use++; 692 rt->u.dst.__use++;
615 skb->dst = (struct dst_entry *) rt; 693
616 return; 694 return rt;
617} 695}
618 696
619struct dst_entry * ip6_route_output(struct sock *sk, struct flowi *fl) 697void ip6_route_input(struct sk_buff *skb)
698{
699 struct ipv6hdr *iph = skb->nh.ipv6h;
700 struct flowi fl = {
701 .iif = skb->dev->ifindex,
702 .nl_u = {
703 .ip6_u = {
704 .daddr = iph->daddr,
705 .saddr = iph->saddr,
706#ifdef CONFIG_IPV6_ROUTE_FWMARK
707 .fwmark = skb->nfmark,
708#endif
709 .flowlabel = (* (u32 *) iph)&IPV6_FLOWINFO_MASK,
710 },
711 },
712 .proto = iph->nexthdr,
713 };
714 int flags = rt6_need_strict(&iph->daddr) ? RT6_LOOKUP_F_IFACE : 0;
715
716 skb->dst = fib6_rule_lookup(&fl, flags, ip6_pol_route_input);
717}
718
719static struct rt6_info *ip6_pol_route_output(struct fib6_table *table,
720 struct flowi *fl, int flags)
620{ 721{
621 struct fib6_node *fn; 722 struct fib6_node *fn;
622 struct rt6_info *rt, *nrt; 723 struct rt6_info *rt, *nrt;
623 int strict; 724 int strict = 0;
624 int attempts = 3; 725 int attempts = 3;
625 int err; 726 int err;
626 int reachable = RT6_SELECT_F_REACHABLE; 727 int reachable = RT6_LOOKUP_F_REACHABLE;
627 728
628 strict = ipv6_addr_type(&fl->fl6_dst) & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL) ? RT6_SELECT_F_IFACE : 0; 729 strict |= flags & RT6_LOOKUP_F_IFACE;
629 730
630relookup: 731relookup:
631 read_lock_bh(&rt6_lock); 732 read_lock_bh(&table->tb6_lock);
632 733
633restart_2: 734restart_2:
634 fn = fib6_lookup(&ip6_routing_table, &fl->fl6_dst, &fl->fl6_src); 735 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
635 736
636restart: 737restart:
637 rt = rt6_select(&fn->leaf, fl->oif, strict | reachable); 738 rt = rt6_select(&fn->leaf, fl->oif, strict | reachable);
638 BACKTRACK(); 739 BACKTRACK(&fl->fl6_src);
639 if (rt == &ip6_null_entry || 740 if (rt == &ip6_null_entry ||
640 rt->rt6i_flags & RTF_CACHE) 741 rt->rt6i_flags & RTF_CACHE)
641 goto out; 742 goto out;
642 743
643 dst_hold(&rt->u.dst); 744 dst_hold(&rt->u.dst);
644 read_unlock_bh(&rt6_lock); 745 read_unlock_bh(&table->tb6_lock);
645 746
646 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP)) 747 if (!rt->rt6i_nexthop && !(rt->rt6i_flags & RTF_NONEXTHOP))
647 nrt = rt6_alloc_cow(rt, &fl->fl6_dst, &fl->fl6_src); 748 nrt = rt6_alloc_cow(rt, &fl->fl6_dst, &fl->fl6_src);
@@ -658,7 +759,7 @@ restart:
658 759
659 dst_hold(&rt->u.dst); 760 dst_hold(&rt->u.dst);
660 if (nrt) { 761 if (nrt) {
661 err = ip6_ins_rt(nrt, NULL, NULL, NULL); 762 err = ip6_ins_rt(nrt);
662 if (!err) 763 if (!err)
663 goto out2; 764 goto out2;
664 } 765 }
@@ -667,7 +768,7 @@ restart:
667 goto out2; 768 goto out2;
668 769
669 /* 770 /*
670 * Race condition! In the gap, when rt6_lock was 771 * Race condition! In the gap, when table->tb6_lock was
671 * released someone could insert this route. Relookup. 772 * released someone could insert this route. Relookup.
672 */ 773 */
673 dst_release(&rt->u.dst); 774 dst_release(&rt->u.dst);
@@ -679,11 +780,21 @@ out:
679 goto restart_2; 780 goto restart_2;
680 } 781 }
681 dst_hold(&rt->u.dst); 782 dst_hold(&rt->u.dst);
682 read_unlock_bh(&rt6_lock); 783 read_unlock_bh(&table->tb6_lock);
683out2: 784out2:
684 rt->u.dst.lastuse = jiffies; 785 rt->u.dst.lastuse = jiffies;
685 rt->u.dst.__use++; 786 rt->u.dst.__use++;
686 return &rt->u.dst; 787 return rt;
788}
789
790struct dst_entry * ip6_route_output(struct sock *sk, struct flowi *fl)
791{
792 int flags = 0;
793
794 if (rt6_need_strict(&fl->fl6_dst))
795 flags |= RT6_LOOKUP_F_IFACE;
796
797 return fib6_rule_lookup(fl, flags, ip6_pol_route_output);
687} 798}
688 799
689 800
@@ -709,7 +820,7 @@ static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
709 820
710 if (rt) { 821 if (rt) {
711 if (rt->rt6i_flags & RTF_CACHE) 822 if (rt->rt6i_flags & RTF_CACHE)
712 ip6_del_rt(rt, NULL, NULL, NULL); 823 ip6_del_rt(rt);
713 else 824 else
714 dst_release(dst); 825 dst_release(dst);
715 } 826 }
@@ -743,11 +854,10 @@ static void ip6_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
743 dst->metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG; 854 dst->metrics[RTAX_FEATURES-1] |= RTAX_FEATURE_ALLFRAG;
744 } 855 }
745 dst->metrics[RTAX_MTU-1] = mtu; 856 dst->metrics[RTAX_MTU-1] = mtu;
857 call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst);
746 } 858 }
747} 859}
748 860
749/* Protected by rt6_lock. */
750static struct dst_entry *ndisc_dst_gc_list;
751static int ipv6_get_mtu(struct net_device *dev); 861static int ipv6_get_mtu(struct net_device *dev);
752 862
753static inline unsigned int ipv6_advmss(unsigned int mtu) 863static inline unsigned int ipv6_advmss(unsigned int mtu)
@@ -768,6 +878,9 @@ static inline unsigned int ipv6_advmss(unsigned int mtu)
768 return mtu; 878 return mtu;
769} 879}
770 880
881static struct dst_entry *ndisc_dst_gc_list;
882static DEFINE_SPINLOCK(ndisc_lock);
883
771struct dst_entry *ndisc_dst_alloc(struct net_device *dev, 884struct dst_entry *ndisc_dst_alloc(struct net_device *dev,
772 struct neighbour *neigh, 885 struct neighbour *neigh,
773 struct in6_addr *addr, 886 struct in6_addr *addr,
@@ -808,10 +921,10 @@ struct dst_entry *ndisc_dst_alloc(struct net_device *dev,
808 rt->rt6i_dst.plen = 128; 921 rt->rt6i_dst.plen = 128;
809#endif 922#endif
810 923
811 write_lock_bh(&rt6_lock); 924 spin_lock_bh(&ndisc_lock);
812 rt->u.dst.next = ndisc_dst_gc_list; 925 rt->u.dst.next = ndisc_dst_gc_list;
813 ndisc_dst_gc_list = &rt->u.dst; 926 ndisc_dst_gc_list = &rt->u.dst;
814 write_unlock_bh(&rt6_lock); 927 spin_unlock_bh(&ndisc_lock);
815 928
816 fib6_force_start_gc(); 929 fib6_force_start_gc();
817 930
@@ -825,8 +938,11 @@ int ndisc_dst_gc(int *more)
825 int freed; 938 int freed;
826 939
827 next = NULL; 940 next = NULL;
941 freed = 0;
942
943 spin_lock_bh(&ndisc_lock);
828 pprev = &ndisc_dst_gc_list; 944 pprev = &ndisc_dst_gc_list;
829 freed = 0; 945
830 while ((dst = *pprev) != NULL) { 946 while ((dst = *pprev) != NULL) {
831 if (!atomic_read(&dst->__refcnt)) { 947 if (!atomic_read(&dst->__refcnt)) {
832 *pprev = dst->next; 948 *pprev = dst->next;
@@ -838,6 +954,8 @@ int ndisc_dst_gc(int *more)
838 } 954 }
839 } 955 }
840 956
957 spin_unlock_bh(&ndisc_lock);
958
841 return freed; 959 return freed;
842} 960}
843 961
@@ -898,28 +1016,24 @@ int ipv6_get_hoplimit(struct net_device *dev)
898 * 1016 *
899 */ 1017 */
900 1018
901int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh, 1019int ip6_route_add(struct fib6_config *cfg)
902 void *_rtattr, struct netlink_skb_parms *req)
903{ 1020{
904 int err; 1021 int err;
905 struct rtmsg *r;
906 struct rtattr **rta;
907 struct rt6_info *rt = NULL; 1022 struct rt6_info *rt = NULL;
908 struct net_device *dev = NULL; 1023 struct net_device *dev = NULL;
909 struct inet6_dev *idev = NULL; 1024 struct inet6_dev *idev = NULL;
1025 struct fib6_table *table;
910 int addr_type; 1026 int addr_type;
911 1027
912 rta = (struct rtattr **) _rtattr; 1028 if (cfg->fc_dst_len > 128 || cfg->fc_src_len > 128)
913
914 if (rtmsg->rtmsg_dst_len > 128 || rtmsg->rtmsg_src_len > 128)
915 return -EINVAL; 1029 return -EINVAL;
916#ifndef CONFIG_IPV6_SUBTREES 1030#ifndef CONFIG_IPV6_SUBTREES
917 if (rtmsg->rtmsg_src_len) 1031 if (cfg->fc_src_len)
918 return -EINVAL; 1032 return -EINVAL;
919#endif 1033#endif
920 if (rtmsg->rtmsg_ifindex) { 1034 if (cfg->fc_ifindex) {
921 err = -ENODEV; 1035 err = -ENODEV;
922 dev = dev_get_by_index(rtmsg->rtmsg_ifindex); 1036 dev = dev_get_by_index(cfg->fc_ifindex);
923 if (!dev) 1037 if (!dev)
924 goto out; 1038 goto out;
925 idev = in6_dev_get(dev); 1039 idev = in6_dev_get(dev);
@@ -927,8 +1041,14 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh,
927 goto out; 1041 goto out;
928 } 1042 }
929 1043
930 if (rtmsg->rtmsg_metric == 0) 1044 if (cfg->fc_metric == 0)
931 rtmsg->rtmsg_metric = IP6_RT_PRIO_USER; 1045 cfg->fc_metric = IP6_RT_PRIO_USER;
1046
1047 table = fib6_new_table(cfg->fc_table);
1048 if (table == NULL) {
1049 err = -ENOBUFS;
1050 goto out;
1051 }
932 1052
933 rt = ip6_dst_alloc(); 1053 rt = ip6_dst_alloc();
934 1054
@@ -938,14 +1058,13 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh,
938 } 1058 }
939 1059
940 rt->u.dst.obsolete = -1; 1060 rt->u.dst.obsolete = -1;
941 rt->rt6i_expires = jiffies + clock_t_to_jiffies(rtmsg->rtmsg_info); 1061 rt->rt6i_expires = jiffies + clock_t_to_jiffies(cfg->fc_expires);
942 if (nlh && (r = NLMSG_DATA(nlh))) {
943 rt->rt6i_protocol = r->rtm_protocol;
944 } else {
945 rt->rt6i_protocol = RTPROT_BOOT;
946 }
947 1062
948 addr_type = ipv6_addr_type(&rtmsg->rtmsg_dst); 1063 if (cfg->fc_protocol == RTPROT_UNSPEC)
1064 cfg->fc_protocol = RTPROT_BOOT;
1065 rt->rt6i_protocol = cfg->fc_protocol;
1066
1067 addr_type = ipv6_addr_type(&cfg->fc_dst);
949 1068
950 if (addr_type & IPV6_ADDR_MULTICAST) 1069 if (addr_type & IPV6_ADDR_MULTICAST)
951 rt->u.dst.input = ip6_mc_input; 1070 rt->u.dst.input = ip6_mc_input;
@@ -954,24 +1073,22 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh,
954 1073
955 rt->u.dst.output = ip6_output; 1074 rt->u.dst.output = ip6_output;
956 1075
957 ipv6_addr_prefix(&rt->rt6i_dst.addr, 1076 ipv6_addr_prefix(&rt->rt6i_dst.addr, &cfg->fc_dst, cfg->fc_dst_len);
958 &rtmsg->rtmsg_dst, rtmsg->rtmsg_dst_len); 1077 rt->rt6i_dst.plen = cfg->fc_dst_len;
959 rt->rt6i_dst.plen = rtmsg->rtmsg_dst_len;
960 if (rt->rt6i_dst.plen == 128) 1078 if (rt->rt6i_dst.plen == 128)
961 rt->u.dst.flags = DST_HOST; 1079 rt->u.dst.flags = DST_HOST;
962 1080
963#ifdef CONFIG_IPV6_SUBTREES 1081#ifdef CONFIG_IPV6_SUBTREES
964 ipv6_addr_prefix(&rt->rt6i_src.addr, 1082 ipv6_addr_prefix(&rt->rt6i_src.addr, &cfg->fc_src, cfg->fc_src_len);
965 &rtmsg->rtmsg_src, rtmsg->rtmsg_src_len); 1083 rt->rt6i_src.plen = cfg->fc_src_len;
966 rt->rt6i_src.plen = rtmsg->rtmsg_src_len;
967#endif 1084#endif
968 1085
969 rt->rt6i_metric = rtmsg->rtmsg_metric; 1086 rt->rt6i_metric = cfg->fc_metric;
970 1087
971 /* We cannot add true routes via loopback here, 1088 /* We cannot add true routes via loopback here,
972 they would result in kernel looping; promote them to reject routes 1089 they would result in kernel looping; promote them to reject routes
973 */ 1090 */
974 if ((rtmsg->rtmsg_flags&RTF_REJECT) || 1091 if ((cfg->fc_flags & RTF_REJECT) ||
975 (dev && (dev->flags&IFF_LOOPBACK) && !(addr_type&IPV6_ADDR_LOOPBACK))) { 1092 (dev && (dev->flags&IFF_LOOPBACK) && !(addr_type&IPV6_ADDR_LOOPBACK))) {
976 /* hold loopback dev/idev if we haven't done so. */ 1093 /* hold loopback dev/idev if we haven't done so. */
977 if (dev != &loopback_dev) { 1094 if (dev != &loopback_dev) {
@@ -994,12 +1111,12 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh,
994 goto install_route; 1111 goto install_route;
995 } 1112 }
996 1113
997 if (rtmsg->rtmsg_flags & RTF_GATEWAY) { 1114 if (cfg->fc_flags & RTF_GATEWAY) {
998 struct in6_addr *gw_addr; 1115 struct in6_addr *gw_addr;
999 int gwa_type; 1116 int gwa_type;
1000 1117
1001 gw_addr = &rtmsg->rtmsg_gateway; 1118 gw_addr = &cfg->fc_gateway;
1002 ipv6_addr_copy(&rt->rt6i_gateway, &rtmsg->rtmsg_gateway); 1119 ipv6_addr_copy(&rt->rt6i_gateway, gw_addr);
1003 gwa_type = ipv6_addr_type(gw_addr); 1120 gwa_type = ipv6_addr_type(gw_addr);
1004 1121
1005 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) { 1122 if (gwa_type != (IPV6_ADDR_LINKLOCAL|IPV6_ADDR_UNICAST)) {
@@ -1016,7 +1133,7 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh,
1016 if (!(gwa_type&IPV6_ADDR_UNICAST)) 1133 if (!(gwa_type&IPV6_ADDR_UNICAST))
1017 goto out; 1134 goto out;
1018 1135
1019 grt = rt6_lookup(gw_addr, NULL, rtmsg->rtmsg_ifindex, 1); 1136 grt = rt6_lookup(gw_addr, NULL, cfg->fc_ifindex, 1);
1020 1137
1021 err = -EHOSTUNREACH; 1138 err = -EHOSTUNREACH;
1022 if (grt == NULL) 1139 if (grt == NULL)
@@ -1048,7 +1165,7 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh,
1048 if (dev == NULL) 1165 if (dev == NULL)
1049 goto out; 1166 goto out;
1050 1167
1051 if (rtmsg->rtmsg_flags & (RTF_GATEWAY|RTF_NONEXTHOP)) { 1168 if (cfg->fc_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) {
1052 rt->rt6i_nexthop = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, dev); 1169 rt->rt6i_nexthop = __neigh_lookup_errno(&nd_tbl, &rt->rt6i_gateway, dev);
1053 if (IS_ERR(rt->rt6i_nexthop)) { 1170 if (IS_ERR(rt->rt6i_nexthop)) {
1054 err = PTR_ERR(rt->rt6i_nexthop); 1171 err = PTR_ERR(rt->rt6i_nexthop);
@@ -1057,24 +1174,24 @@ int ip6_route_add(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh,
1057 } 1174 }
1058 } 1175 }
1059 1176
1060 rt->rt6i_flags = rtmsg->rtmsg_flags; 1177 rt->rt6i_flags = cfg->fc_flags;
1061 1178
1062install_route: 1179install_route:
1063 if (rta && rta[RTA_METRICS-1]) { 1180 if (cfg->fc_mx) {
1064 int attrlen = RTA_PAYLOAD(rta[RTA_METRICS-1]); 1181 struct nlattr *nla;
1065 struct rtattr *attr = RTA_DATA(rta[RTA_METRICS-1]); 1182 int remaining;
1066 1183
1067 while (RTA_OK(attr, attrlen)) { 1184 nla_for_each_attr(nla, cfg->fc_mx, cfg->fc_mx_len, remaining) {
1068 unsigned flavor = attr->rta_type; 1185 int type = nla->nla_type;
1069 if (flavor) { 1186
1070 if (flavor > RTAX_MAX) { 1187 if (type) {
1188 if (type > RTAX_MAX) {
1071 err = -EINVAL; 1189 err = -EINVAL;
1072 goto out; 1190 goto out;
1073 } 1191 }
1074 rt->u.dst.metrics[flavor-1] = 1192
1075 *(u32 *)RTA_DATA(attr); 1193 rt->u.dst.metrics[type - 1] = nla_get_u32(nla);
1076 } 1194 }
1077 attr = RTA_NEXT(attr, attrlen);
1078 } 1195 }
1079 } 1196 }
1080 1197
@@ -1086,7 +1203,8 @@ install_route:
1086 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(dst_mtu(&rt->u.dst)); 1203 rt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(dst_mtu(&rt->u.dst));
1087 rt->u.dst.dev = dev; 1204 rt->u.dst.dev = dev;
1088 rt->rt6i_idev = idev; 1205 rt->rt6i_idev = idev;
1089 return ip6_ins_rt(rt, nlh, _rtattr, req); 1206 rt->rt6i_table = table;
1207 return __ip6_ins_rt(rt, &cfg->fc_nlinfo);
1090 1208
1091out: 1209out:
1092 if (dev) 1210 if (dev)
@@ -1098,51 +1216,65 @@ out:
1098 return err; 1216 return err;
1099} 1217}
1100 1218
1101int ip6_del_rt(struct rt6_info *rt, struct nlmsghdr *nlh, void *_rtattr, struct netlink_skb_parms *req) 1219static int __ip6_del_rt(struct rt6_info *rt, struct nl_info *info)
1102{ 1220{
1103 int err; 1221 int err;
1222 struct fib6_table *table;
1104 1223
1105 write_lock_bh(&rt6_lock); 1224 if (rt == &ip6_null_entry)
1225 return -ENOENT;
1106 1226
1107 err = fib6_del(rt, nlh, _rtattr, req); 1227 table = rt->rt6i_table;
1228 write_lock_bh(&table->tb6_lock);
1229
1230 err = fib6_del(rt, info);
1108 dst_release(&rt->u.dst); 1231 dst_release(&rt->u.dst);
1109 1232
1110 write_unlock_bh(&rt6_lock); 1233 write_unlock_bh(&table->tb6_lock);
1111 1234
1112 return err; 1235 return err;
1113} 1236}
1114 1237
1115static int ip6_route_del(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh, void *_rtattr, struct netlink_skb_parms *req) 1238int ip6_del_rt(struct rt6_info *rt)
1116{ 1239{
1240 return __ip6_del_rt(rt, NULL);
1241}
1242
1243static int ip6_route_del(struct fib6_config *cfg)
1244{
1245 struct fib6_table *table;
1117 struct fib6_node *fn; 1246 struct fib6_node *fn;
1118 struct rt6_info *rt; 1247 struct rt6_info *rt;
1119 int err = -ESRCH; 1248 int err = -ESRCH;
1120 1249
1121 read_lock_bh(&rt6_lock); 1250 table = fib6_get_table(cfg->fc_table);
1251 if (table == NULL)
1252 return err;
1122 1253
1123 fn = fib6_locate(&ip6_routing_table, 1254 read_lock_bh(&table->tb6_lock);
1124 &rtmsg->rtmsg_dst, rtmsg->rtmsg_dst_len, 1255
1125 &rtmsg->rtmsg_src, rtmsg->rtmsg_src_len); 1256 fn = fib6_locate(&table->tb6_root,
1257 &cfg->fc_dst, cfg->fc_dst_len,
1258 &cfg->fc_src, cfg->fc_src_len);
1126 1259
1127 if (fn) { 1260 if (fn) {
1128 for (rt = fn->leaf; rt; rt = rt->u.next) { 1261 for (rt = fn->leaf; rt; rt = rt->u.next) {
1129 if (rtmsg->rtmsg_ifindex && 1262 if (cfg->fc_ifindex &&
1130 (rt->rt6i_dev == NULL || 1263 (rt->rt6i_dev == NULL ||
1131 rt->rt6i_dev->ifindex != rtmsg->rtmsg_ifindex)) 1264 rt->rt6i_dev->ifindex != cfg->fc_ifindex))
1132 continue; 1265 continue;
1133 if (rtmsg->rtmsg_flags&RTF_GATEWAY && 1266 if (cfg->fc_flags & RTF_GATEWAY &&
1134 !ipv6_addr_equal(&rtmsg->rtmsg_gateway, &rt->rt6i_gateway)) 1267 !ipv6_addr_equal(&cfg->fc_gateway, &rt->rt6i_gateway))
1135 continue; 1268 continue;
1136 if (rtmsg->rtmsg_metric && 1269 if (cfg->fc_metric && cfg->fc_metric != rt->rt6i_metric)
1137 rtmsg->rtmsg_metric != rt->rt6i_metric)
1138 continue; 1270 continue;
1139 dst_hold(&rt->u.dst); 1271 dst_hold(&rt->u.dst);
1140 read_unlock_bh(&rt6_lock); 1272 read_unlock_bh(&table->tb6_lock);
1141 1273
1142 return ip6_del_rt(rt, nlh, _rtattr, req); 1274 return __ip6_del_rt(rt, &cfg->fc_nlinfo);
1143 } 1275 }
1144 } 1276 }
1145 read_unlock_bh(&rt6_lock); 1277 read_unlock_bh(&table->tb6_lock);
1146 1278
1147 return err; 1279 return err;
1148} 1280}
@@ -1150,11 +1282,17 @@ static int ip6_route_del(struct in6_rtmsg *rtmsg, struct nlmsghdr *nlh, void *_r
1150/* 1282/*
1151 * Handle redirects 1283 * Handle redirects
1152 */ 1284 */
1153void rt6_redirect(struct in6_addr *dest, struct in6_addr *saddr, 1285struct ip6rd_flowi {
1154 struct neighbour *neigh, u8 *lladdr, int on_link) 1286 struct flowi fl;
1287 struct in6_addr gateway;
1288};
1289
1290static struct rt6_info *__ip6_route_redirect(struct fib6_table *table,
1291 struct flowi *fl,
1292 int flags)
1155{ 1293{
1156 struct rt6_info *rt, *nrt = NULL; 1294 struct ip6rd_flowi *rdfl = (struct ip6rd_flowi *)fl;
1157 int strict; 1295 struct rt6_info *rt;
1158 struct fib6_node *fn; 1296 struct fib6_node *fn;
1159 1297
1160 /* 1298 /*
@@ -1167,10 +1305,9 @@ void rt6_redirect(struct in6_addr *dest, struct in6_addr *saddr,
1167 * is a bit fuzzy and one might need to check all possible 1305 * is a bit fuzzy and one might need to check all possible
1168 * routes. 1306 * routes.
1169 */ 1307 */
1170 strict = ipv6_addr_type(dest) & (IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL);
1171 1308
1172 read_lock_bh(&rt6_lock); 1309 read_lock_bh(&table->tb6_lock);
1173 fn = fib6_lookup(&ip6_routing_table, dest, NULL); 1310 fn = fib6_lookup(&table->tb6_root, &fl->fl6_dst, &fl->fl6_src);
1174restart: 1311restart:
1175 for (rt = fn->leaf; rt; rt = rt->u.next) { 1312 for (rt = fn->leaf; rt; rt = rt->u.next) {
1176 /* 1313 /*
@@ -1185,29 +1322,60 @@ restart:
1185 continue; 1322 continue;
1186 if (!(rt->rt6i_flags & RTF_GATEWAY)) 1323 if (!(rt->rt6i_flags & RTF_GATEWAY))
1187 continue; 1324 continue;
1188 if (neigh->dev != rt->rt6i_dev) 1325 if (fl->oif != rt->rt6i_dev->ifindex)
1189 continue; 1326 continue;
1190 if (!ipv6_addr_equal(saddr, &rt->rt6i_gateway)) 1327 if (!ipv6_addr_equal(&rdfl->gateway, &rt->rt6i_gateway))
1191 continue; 1328 continue;
1192 break; 1329 break;
1193 } 1330 }
1194 if (rt)
1195 dst_hold(&rt->u.dst);
1196 else if (strict) {
1197 while ((fn = fn->parent) != NULL) {
1198 if (fn->fn_flags & RTN_ROOT)
1199 break;
1200 if (fn->fn_flags & RTN_RTINFO)
1201 goto restart;
1202 }
1203 }
1204 read_unlock_bh(&rt6_lock);
1205 1331
1206 if (!rt) { 1332 if (!rt)
1333 rt = &ip6_null_entry;
1334 BACKTRACK(&fl->fl6_src);
1335out:
1336 dst_hold(&rt->u.dst);
1337
1338 read_unlock_bh(&table->tb6_lock);
1339
1340 return rt;
1341};
1342
1343static struct rt6_info *ip6_route_redirect(struct in6_addr *dest,
1344 struct in6_addr *src,
1345 struct in6_addr *gateway,
1346 struct net_device *dev)
1347{
1348 struct ip6rd_flowi rdfl = {
1349 .fl = {
1350 .oif = dev->ifindex,
1351 .nl_u = {
1352 .ip6_u = {
1353 .daddr = *dest,
1354 .saddr = *src,
1355 },
1356 },
1357 },
1358 .gateway = *gateway,
1359 };
1360 int flags = rt6_need_strict(dest) ? RT6_LOOKUP_F_IFACE : 0;
1361
1362 return (struct rt6_info *)fib6_rule_lookup((struct flowi *)&rdfl, flags, __ip6_route_redirect);
1363}
1364
1365void rt6_redirect(struct in6_addr *dest, struct in6_addr *src,
1366 struct in6_addr *saddr,
1367 struct neighbour *neigh, u8 *lladdr, int on_link)
1368{
1369 struct rt6_info *rt, *nrt = NULL;
1370 struct netevent_redirect netevent;
1371
1372 rt = ip6_route_redirect(dest, src, saddr, neigh->dev);
1373
1374 if (rt == &ip6_null_entry) {
1207 if (net_ratelimit()) 1375 if (net_ratelimit())
1208 printk(KERN_DEBUG "rt6_redirect: source isn't a valid nexthop " 1376 printk(KERN_DEBUG "rt6_redirect: source isn't a valid nexthop "
1209 "for redirect target\n"); 1377 "for redirect target\n");
1210 return; 1378 goto out;
1211 } 1379 }
1212 1380
1213 /* 1381 /*
@@ -1250,11 +1418,15 @@ restart:
1250 nrt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(neigh->dev); 1418 nrt->u.dst.metrics[RTAX_MTU-1] = ipv6_get_mtu(neigh->dev);
1251 nrt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(dst_mtu(&nrt->u.dst)); 1419 nrt->u.dst.metrics[RTAX_ADVMSS-1] = ipv6_advmss(dst_mtu(&nrt->u.dst));
1252 1420
1253 if (ip6_ins_rt(nrt, NULL, NULL, NULL)) 1421 if (ip6_ins_rt(nrt))
1254 goto out; 1422 goto out;
1255 1423
1424 netevent.old = &rt->u.dst;
1425 netevent.new = &nrt->u.dst;
1426 call_netevent_notifiers(NETEVENT_REDIRECT, &netevent);
1427
1256 if (rt->rt6i_flags&RTF_CACHE) { 1428 if (rt->rt6i_flags&RTF_CACHE) {
1257 ip6_del_rt(rt, NULL, NULL, NULL); 1429 ip6_del_rt(rt);
1258 return; 1430 return;
1259 } 1431 }
1260 1432
@@ -1336,7 +1508,7 @@ void rt6_pmtu_discovery(struct in6_addr *daddr, struct in6_addr *saddr,
1336 dst_set_expires(&nrt->u.dst, ip6_rt_mtu_expires); 1508 dst_set_expires(&nrt->u.dst, ip6_rt_mtu_expires);
1337 nrt->rt6i_flags |= RTF_DYNAMIC|RTF_EXPIRES; 1509 nrt->rt6i_flags |= RTF_DYNAMIC|RTF_EXPIRES;
1338 1510
1339 ip6_ins_rt(nrt, NULL, NULL, NULL); 1511 ip6_ins_rt(nrt);
1340 } 1512 }
1341out: 1513out:
1342 dst_release(&rt->u.dst); 1514 dst_release(&rt->u.dst);
@@ -1372,6 +1544,7 @@ static struct rt6_info * ip6_rt_copy(struct rt6_info *ort)
1372#ifdef CONFIG_IPV6_SUBTREES 1544#ifdef CONFIG_IPV6_SUBTREES
1373 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key)); 1545 memcpy(&rt->rt6i_src, &ort->rt6i_src, sizeof(struct rt6key));
1374#endif 1546#endif
1547 rt->rt6i_table = ort->rt6i_table;
1375 } 1548 }
1376 return rt; 1549 return rt;
1377} 1550}
@@ -1382,9 +1555,14 @@ static struct rt6_info *rt6_get_route_info(struct in6_addr *prefix, int prefixle
1382{ 1555{
1383 struct fib6_node *fn; 1556 struct fib6_node *fn;
1384 struct rt6_info *rt = NULL; 1557 struct rt6_info *rt = NULL;
1558 struct fib6_table *table;
1385 1559
1386 write_lock_bh(&rt6_lock); 1560 table = fib6_get_table(RT6_TABLE_INFO);
1387 fn = fib6_locate(&ip6_routing_table, prefix ,prefixlen, NULL, 0); 1561 if (table == NULL)
1562 return NULL;
1563
1564 write_lock_bh(&table->tb6_lock);
1565 fn = fib6_locate(&table->tb6_root, prefix ,prefixlen, NULL, 0);
1388 if (!fn) 1566 if (!fn)
1389 goto out; 1567 goto out;
1390 1568
@@ -1399,7 +1577,7 @@ static struct rt6_info *rt6_get_route_info(struct in6_addr *prefix, int prefixle
1399 break; 1577 break;
1400 } 1578 }
1401out: 1579out:
1402 write_unlock_bh(&rt6_lock); 1580 write_unlock_bh(&table->tb6_lock);
1403 return rt; 1581 return rt;
1404} 1582}
1405 1583
@@ -1407,21 +1585,23 @@ static struct rt6_info *rt6_add_route_info(struct in6_addr *prefix, int prefixle
1407 struct in6_addr *gwaddr, int ifindex, 1585 struct in6_addr *gwaddr, int ifindex,
1408 unsigned pref) 1586 unsigned pref)
1409{ 1587{
1410 struct in6_rtmsg rtmsg; 1588 struct fib6_config cfg = {
1589 .fc_table = RT6_TABLE_INFO,
1590 .fc_metric = 1024,
1591 .fc_ifindex = ifindex,
1592 .fc_dst_len = prefixlen,
1593 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO |
1594 RTF_UP | RTF_PREF(pref),
1595 };
1596
1597 ipv6_addr_copy(&cfg.fc_dst, prefix);
1598 ipv6_addr_copy(&cfg.fc_gateway, gwaddr);
1411 1599
1412 memset(&rtmsg, 0, sizeof(rtmsg));
1413 rtmsg.rtmsg_type = RTMSG_NEWROUTE;
1414 ipv6_addr_copy(&rtmsg.rtmsg_dst, prefix);
1415 rtmsg.rtmsg_dst_len = prefixlen;
1416 ipv6_addr_copy(&rtmsg.rtmsg_gateway, gwaddr);
1417 rtmsg.rtmsg_metric = 1024;
1418 rtmsg.rtmsg_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_ROUTEINFO | RTF_UP | RTF_PREF(pref);
1419 /* We should treat it as a default route if prefix length is 0. */ 1600 /* We should treat it as a default route if prefix length is 0. */
1420 if (!prefixlen) 1601 if (!prefixlen)
1421 rtmsg.rtmsg_flags |= RTF_DEFAULT; 1602 cfg.fc_flags |= RTF_DEFAULT;
1422 rtmsg.rtmsg_ifindex = ifindex;
1423 1603
1424 ip6_route_add(&rtmsg, NULL, NULL, NULL); 1604 ip6_route_add(&cfg);
1425 1605
1426 return rt6_get_route_info(prefix, prefixlen, gwaddr, ifindex); 1606 return rt6_get_route_info(prefix, prefixlen, gwaddr, ifindex);
1427} 1607}
@@ -1430,12 +1610,14 @@ static struct rt6_info *rt6_add_route_info(struct in6_addr *prefix, int prefixle
1430struct rt6_info *rt6_get_dflt_router(struct in6_addr *addr, struct net_device *dev) 1610struct rt6_info *rt6_get_dflt_router(struct in6_addr *addr, struct net_device *dev)
1431{ 1611{
1432 struct rt6_info *rt; 1612 struct rt6_info *rt;
1433 struct fib6_node *fn; 1613 struct fib6_table *table;
1434 1614
1435 fn = &ip6_routing_table; 1615 table = fib6_get_table(RT6_TABLE_DFLT);
1616 if (table == NULL)
1617 return NULL;
1436 1618
1437 write_lock_bh(&rt6_lock); 1619 write_lock_bh(&table->tb6_lock);
1438 for (rt = fn->leaf; rt; rt=rt->u.next) { 1620 for (rt = table->tb6_root.leaf; rt; rt=rt->u.next) {
1439 if (dev == rt->rt6i_dev && 1621 if (dev == rt->rt6i_dev &&
1440 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) && 1622 ((rt->rt6i_flags & (RTF_ADDRCONF | RTF_DEFAULT)) == (RTF_ADDRCONF | RTF_DEFAULT)) &&
1441 ipv6_addr_equal(&rt->rt6i_gateway, addr)) 1623 ipv6_addr_equal(&rt->rt6i_gateway, addr))
@@ -1443,7 +1625,7 @@ struct rt6_info *rt6_get_dflt_router(struct in6_addr *addr, struct net_device *d
1443 } 1625 }
1444 if (rt) 1626 if (rt)
1445 dst_hold(&rt->u.dst); 1627 dst_hold(&rt->u.dst);
1446 write_unlock_bh(&rt6_lock); 1628 write_unlock_bh(&table->tb6_lock);
1447 return rt; 1629 return rt;
1448} 1630}
1449 1631
@@ -1451,43 +1633,65 @@ struct rt6_info *rt6_add_dflt_router(struct in6_addr *gwaddr,
1451 struct net_device *dev, 1633 struct net_device *dev,
1452 unsigned int pref) 1634 unsigned int pref)
1453{ 1635{
1454 struct in6_rtmsg rtmsg; 1636 struct fib6_config cfg = {
1637 .fc_table = RT6_TABLE_DFLT,
1638 .fc_metric = 1024,
1639 .fc_ifindex = dev->ifindex,
1640 .fc_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT |
1641 RTF_UP | RTF_EXPIRES | RTF_PREF(pref),
1642 };
1455 1643
1456 memset(&rtmsg, 0, sizeof(struct in6_rtmsg)); 1644 ipv6_addr_copy(&cfg.fc_gateway, gwaddr);
1457 rtmsg.rtmsg_type = RTMSG_NEWROUTE;
1458 ipv6_addr_copy(&rtmsg.rtmsg_gateway, gwaddr);
1459 rtmsg.rtmsg_metric = 1024;
1460 rtmsg.rtmsg_flags = RTF_GATEWAY | RTF_ADDRCONF | RTF_DEFAULT | RTF_UP | RTF_EXPIRES |
1461 RTF_PREF(pref);
1462 1645
1463 rtmsg.rtmsg_ifindex = dev->ifindex; 1646 ip6_route_add(&cfg);
1464 1647
1465 ip6_route_add(&rtmsg, NULL, NULL, NULL);
1466 return rt6_get_dflt_router(gwaddr, dev); 1648 return rt6_get_dflt_router(gwaddr, dev);
1467} 1649}
1468 1650
1469void rt6_purge_dflt_routers(void) 1651void rt6_purge_dflt_routers(void)
1470{ 1652{
1471 struct rt6_info *rt; 1653 struct rt6_info *rt;
1654 struct fib6_table *table;
1655
1656 /* NOTE: Keep consistent with rt6_get_dflt_router */
1657 table = fib6_get_table(RT6_TABLE_DFLT);
1658 if (table == NULL)
1659 return;
1472 1660
1473restart: 1661restart:
1474 read_lock_bh(&rt6_lock); 1662 read_lock_bh(&table->tb6_lock);
1475 for (rt = ip6_routing_table.leaf; rt; rt = rt->u.next) { 1663 for (rt = table->tb6_root.leaf; rt; rt = rt->u.next) {
1476 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) { 1664 if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) {
1477 dst_hold(&rt->u.dst); 1665 dst_hold(&rt->u.dst);
1478 1666 read_unlock_bh(&table->tb6_lock);
1479 read_unlock_bh(&rt6_lock); 1667 ip6_del_rt(rt);
1480
1481 ip6_del_rt(rt, NULL, NULL, NULL);
1482
1483 goto restart; 1668 goto restart;
1484 } 1669 }
1485 } 1670 }
1486 read_unlock_bh(&rt6_lock); 1671 read_unlock_bh(&table->tb6_lock);
1672}
1673
1674static void rtmsg_to_fib6_config(struct in6_rtmsg *rtmsg,
1675 struct fib6_config *cfg)
1676{
1677 memset(cfg, 0, sizeof(*cfg));
1678
1679 cfg->fc_table = RT6_TABLE_MAIN;
1680 cfg->fc_ifindex = rtmsg->rtmsg_ifindex;
1681 cfg->fc_metric = rtmsg->rtmsg_metric;
1682 cfg->fc_expires = rtmsg->rtmsg_info;
1683 cfg->fc_dst_len = rtmsg->rtmsg_dst_len;
1684 cfg->fc_src_len = rtmsg->rtmsg_src_len;
1685 cfg->fc_flags = rtmsg->rtmsg_flags;
1686
1687 ipv6_addr_copy(&cfg->fc_dst, &rtmsg->rtmsg_dst);
1688 ipv6_addr_copy(&cfg->fc_src, &rtmsg->rtmsg_src);
1689 ipv6_addr_copy(&cfg->fc_gateway, &rtmsg->rtmsg_gateway);
1487} 1690}
1488 1691
1489int ipv6_route_ioctl(unsigned int cmd, void __user *arg) 1692int ipv6_route_ioctl(unsigned int cmd, void __user *arg)
1490{ 1693{
1694 struct fib6_config cfg;
1491 struct in6_rtmsg rtmsg; 1695 struct in6_rtmsg rtmsg;
1492 int err; 1696 int err;
1493 1697
@@ -1500,14 +1704,16 @@ int ipv6_route_ioctl(unsigned int cmd, void __user *arg)
1500 sizeof(struct in6_rtmsg)); 1704 sizeof(struct in6_rtmsg));
1501 if (err) 1705 if (err)
1502 return -EFAULT; 1706 return -EFAULT;
1503 1707
1708 rtmsg_to_fib6_config(&rtmsg, &cfg);
1709
1504 rtnl_lock(); 1710 rtnl_lock();
1505 switch (cmd) { 1711 switch (cmd) {
1506 case SIOCADDRT: 1712 case SIOCADDRT:
1507 err = ip6_route_add(&rtmsg, NULL, NULL, NULL); 1713 err = ip6_route_add(&cfg);
1508 break; 1714 break;
1509 case SIOCDELRT: 1715 case SIOCDELRT:
1510 err = ip6_route_del(&rtmsg, NULL, NULL, NULL); 1716 err = ip6_route_del(&cfg);
1511 break; 1717 break;
1512 default: 1718 default:
1513 err = -EINVAL; 1719 err = -EINVAL;
@@ -1526,6 +1732,10 @@ int ipv6_route_ioctl(unsigned int cmd, void __user *arg)
1526 1732
1527static int ip6_pkt_discard(struct sk_buff *skb) 1733static int ip6_pkt_discard(struct sk_buff *skb)
1528{ 1734{
1735 int type = ipv6_addr_type(&skb->nh.ipv6h->daddr);
1736 if (type == IPV6_ADDR_ANY || type == IPV6_ADDR_RESERVED)
1737 IP6_INC_STATS(IPSTATS_MIB_INADDRERRORS);
1738
1529 IP6_INC_STATS(IPSTATS_MIB_OUTNOROUTES); 1739 IP6_INC_STATS(IPSTATS_MIB_OUTNOROUTES);
1530 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_NOROUTE, 0, skb->dev); 1740 icmpv6_send(skb, ICMPV6_DEST_UNREACH, ICMPV6_NOROUTE, 0, skb->dev);
1531 kfree_skb(skb); 1741 kfree_skb(skb);
@@ -1577,6 +1787,7 @@ struct rt6_info *addrconf_dst_alloc(struct inet6_dev *idev,
1577 1787
1578 ipv6_addr_copy(&rt->rt6i_dst.addr, addr); 1788 ipv6_addr_copy(&rt->rt6i_dst.addr, addr);
1579 rt->rt6i_dst.plen = 128; 1789 rt->rt6i_dst.plen = 128;
1790 rt->rt6i_table = fib6_get_table(RT6_TABLE_LOCAL);
1580 1791
1581 atomic_set(&rt->u.dst.__refcnt, 1); 1792 atomic_set(&rt->u.dst.__refcnt, 1);
1582 1793
@@ -1595,9 +1806,7 @@ static int fib6_ifdown(struct rt6_info *rt, void *arg)
1595 1806
1596void rt6_ifdown(struct net_device *dev) 1807void rt6_ifdown(struct net_device *dev)
1597{ 1808{
1598 write_lock_bh(&rt6_lock); 1809 fib6_clean_all(fib6_ifdown, 0, dev);
1599 fib6_clean_tree(&ip6_routing_table, fib6_ifdown, 0, dev);
1600 write_unlock_bh(&rt6_lock);
1601} 1810}
1602 1811
1603struct rt6_mtu_change_arg 1812struct rt6_mtu_change_arg
@@ -1647,80 +1856,114 @@ static int rt6_mtu_change_route(struct rt6_info *rt, void *p_arg)
1647 1856
1648void rt6_mtu_change(struct net_device *dev, unsigned mtu) 1857void rt6_mtu_change(struct net_device *dev, unsigned mtu)
1649{ 1858{
1650 struct rt6_mtu_change_arg arg; 1859 struct rt6_mtu_change_arg arg = {
1860 .dev = dev,
1861 .mtu = mtu,
1862 };
1651 1863
1652 arg.dev = dev; 1864 fib6_clean_all(rt6_mtu_change_route, 0, &arg);
1653 arg.mtu = mtu;
1654 read_lock_bh(&rt6_lock);
1655 fib6_clean_tree(&ip6_routing_table, rt6_mtu_change_route, 0, &arg);
1656 read_unlock_bh(&rt6_lock);
1657} 1865}
1658 1866
1659static int inet6_rtm_to_rtmsg(struct rtmsg *r, struct rtattr **rta, 1867static struct nla_policy rtm_ipv6_policy[RTA_MAX+1] __read_mostly = {
1660 struct in6_rtmsg *rtmsg) 1868 [RTA_GATEWAY] = { .len = sizeof(struct in6_addr) },
1869 [RTA_OIF] = { .type = NLA_U32 },
1870 [RTA_IIF] = { .type = NLA_U32 },
1871 [RTA_PRIORITY] = { .type = NLA_U32 },
1872 [RTA_METRICS] = { .type = NLA_NESTED },
1873};
1874
1875static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
1876 struct fib6_config *cfg)
1661{ 1877{
1662 memset(rtmsg, 0, sizeof(*rtmsg)); 1878 struct rtmsg *rtm;
1879 struct nlattr *tb[RTA_MAX+1];
1880 int err;
1663 1881
1664 rtmsg->rtmsg_dst_len = r->rtm_dst_len; 1882 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
1665 rtmsg->rtmsg_src_len = r->rtm_src_len; 1883 if (err < 0)
1666 rtmsg->rtmsg_flags = RTF_UP; 1884 goto errout;
1667 if (r->rtm_type == RTN_UNREACHABLE)
1668 rtmsg->rtmsg_flags |= RTF_REJECT;
1669 1885
1670 if (rta[RTA_GATEWAY-1]) { 1886 err = -EINVAL;
1671 if (rta[RTA_GATEWAY-1]->rta_len != RTA_LENGTH(16)) 1887 rtm = nlmsg_data(nlh);
1672 return -EINVAL; 1888 memset(cfg, 0, sizeof(*cfg));
1673 memcpy(&rtmsg->rtmsg_gateway, RTA_DATA(rta[RTA_GATEWAY-1]), 16); 1889
1674 rtmsg->rtmsg_flags |= RTF_GATEWAY; 1890 cfg->fc_table = rtm->rtm_table;
1675 } 1891 cfg->fc_dst_len = rtm->rtm_dst_len;
1676 if (rta[RTA_DST-1]) { 1892 cfg->fc_src_len = rtm->rtm_src_len;
1677 if (RTA_PAYLOAD(rta[RTA_DST-1]) < ((r->rtm_dst_len+7)>>3)) 1893 cfg->fc_flags = RTF_UP;
1678 return -EINVAL; 1894 cfg->fc_protocol = rtm->rtm_protocol;
1679 memcpy(&rtmsg->rtmsg_dst, RTA_DATA(rta[RTA_DST-1]), ((r->rtm_dst_len+7)>>3)); 1895
1896 if (rtm->rtm_type == RTN_UNREACHABLE)
1897 cfg->fc_flags |= RTF_REJECT;
1898
1899 cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid;
1900 cfg->fc_nlinfo.nlh = nlh;
1901
1902 if (tb[RTA_GATEWAY]) {
1903 nla_memcpy(&cfg->fc_gateway, tb[RTA_GATEWAY], 16);
1904 cfg->fc_flags |= RTF_GATEWAY;
1680 } 1905 }
1681 if (rta[RTA_SRC-1]) { 1906
1682 if (RTA_PAYLOAD(rta[RTA_SRC-1]) < ((r->rtm_src_len+7)>>3)) 1907 if (tb[RTA_DST]) {
1683 return -EINVAL; 1908 int plen = (rtm->rtm_dst_len + 7) >> 3;
1684 memcpy(&rtmsg->rtmsg_src, RTA_DATA(rta[RTA_SRC-1]), ((r->rtm_src_len+7)>>3)); 1909
1910 if (nla_len(tb[RTA_DST]) < plen)
1911 goto errout;
1912
1913 nla_memcpy(&cfg->fc_dst, tb[RTA_DST], plen);
1685 } 1914 }
1686 if (rta[RTA_OIF-1]) { 1915
1687 if (rta[RTA_OIF-1]->rta_len != RTA_LENGTH(sizeof(int))) 1916 if (tb[RTA_SRC]) {
1688 return -EINVAL; 1917 int plen = (rtm->rtm_src_len + 7) >> 3;
1689 memcpy(&rtmsg->rtmsg_ifindex, RTA_DATA(rta[RTA_OIF-1]), sizeof(int)); 1918
1919 if (nla_len(tb[RTA_SRC]) < plen)
1920 goto errout;
1921
1922 nla_memcpy(&cfg->fc_src, tb[RTA_SRC], plen);
1690 } 1923 }
1691 if (rta[RTA_PRIORITY-1]) { 1924
1692 if (rta[RTA_PRIORITY-1]->rta_len != RTA_LENGTH(4)) 1925 if (tb[RTA_OIF])
1693 return -EINVAL; 1926 cfg->fc_ifindex = nla_get_u32(tb[RTA_OIF]);
1694 memcpy(&rtmsg->rtmsg_metric, RTA_DATA(rta[RTA_PRIORITY-1]), 4); 1927
1928 if (tb[RTA_PRIORITY])
1929 cfg->fc_metric = nla_get_u32(tb[RTA_PRIORITY]);
1930
1931 if (tb[RTA_METRICS]) {
1932 cfg->fc_mx = nla_data(tb[RTA_METRICS]);
1933 cfg->fc_mx_len = nla_len(tb[RTA_METRICS]);
1695 } 1934 }
1696 return 0; 1935
1936 if (tb[RTA_TABLE])
1937 cfg->fc_table = nla_get_u32(tb[RTA_TABLE]);
1938
1939 err = 0;
1940errout:
1941 return err;
1697} 1942}
1698 1943
1699int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) 1944int inet6_rtm_delroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
1700{ 1945{
1701 struct rtmsg *r = NLMSG_DATA(nlh); 1946 struct fib6_config cfg;
1702 struct in6_rtmsg rtmsg; 1947 int err;
1703 1948
1704 if (inet6_rtm_to_rtmsg(r, arg, &rtmsg)) 1949 err = rtm_to_fib6_config(skb, nlh, &cfg);
1705 return -EINVAL; 1950 if (err < 0)
1706 return ip6_route_del(&rtmsg, nlh, arg, &NETLINK_CB(skb)); 1951 return err;
1952
1953 return ip6_route_del(&cfg);
1707} 1954}
1708 1955
1709int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg) 1956int inet6_rtm_newroute(struct sk_buff *skb, struct nlmsghdr* nlh, void *arg)
1710{ 1957{
1711 struct rtmsg *r = NLMSG_DATA(nlh); 1958 struct fib6_config cfg;
1712 struct in6_rtmsg rtmsg; 1959 int err;
1713 1960
1714 if (inet6_rtm_to_rtmsg(r, arg, &rtmsg)) 1961 err = rtm_to_fib6_config(skb, nlh, &cfg);
1715 return -EINVAL; 1962 if (err < 0)
1716 return ip6_route_add(&rtmsg, nlh, arg, &NETLINK_CB(skb)); 1963 return err;
1717}
1718 1964
1719struct rt6_rtnl_dump_arg 1965 return ip6_route_add(&cfg);
1720{ 1966}
1721 struct sk_buff *skb;
1722 struct netlink_callback *cb;
1723};
1724 1967
1725static int rt6_fill_node(struct sk_buff *skb, struct rt6_info *rt, 1968static int rt6_fill_node(struct sk_buff *skb, struct rt6_info *rt,
1726 struct in6_addr *dst, struct in6_addr *src, 1969 struct in6_addr *dst, struct in6_addr *src,
@@ -1728,9 +1971,9 @@ static int rt6_fill_node(struct sk_buff *skb, struct rt6_info *rt,
1728 int prefix, unsigned int flags) 1971 int prefix, unsigned int flags)
1729{ 1972{
1730 struct rtmsg *rtm; 1973 struct rtmsg *rtm;
1731 struct nlmsghdr *nlh; 1974 struct nlmsghdr *nlh;
1732 unsigned char *b = skb->tail;
1733 struct rta_cacheinfo ci; 1975 struct rta_cacheinfo ci;
1976 u32 table;
1734 1977
1735 if (prefix) { /* user wants prefix routes only */ 1978 if (prefix) { /* user wants prefix routes only */
1736 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) { 1979 if (!(rt->rt6i_flags & RTF_PREFIX_RT)) {
@@ -1739,13 +1982,21 @@ static int rt6_fill_node(struct sk_buff *skb, struct rt6_info *rt,
1739 } 1982 }
1740 } 1983 }
1741 1984
1742 nlh = NLMSG_NEW(skb, pid, seq, type, sizeof(*rtm), flags); 1985 nlh = nlmsg_put(skb, pid, seq, type, sizeof(*rtm), flags);
1743 rtm = NLMSG_DATA(nlh); 1986 if (nlh == NULL)
1987 return -ENOBUFS;
1988
1989 rtm = nlmsg_data(nlh);
1744 rtm->rtm_family = AF_INET6; 1990 rtm->rtm_family = AF_INET6;
1745 rtm->rtm_dst_len = rt->rt6i_dst.plen; 1991 rtm->rtm_dst_len = rt->rt6i_dst.plen;
1746 rtm->rtm_src_len = rt->rt6i_src.plen; 1992 rtm->rtm_src_len = rt->rt6i_src.plen;
1747 rtm->rtm_tos = 0; 1993 rtm->rtm_tos = 0;
1748 rtm->rtm_table = RT_TABLE_MAIN; 1994 if (rt->rt6i_table)
1995 table = rt->rt6i_table->tb6_id;
1996 else
1997 table = RT6_TABLE_UNSPEC;
1998 rtm->rtm_table = table;
1999 NLA_PUT_U32(skb, RTA_TABLE, table);
1749 if (rt->rt6i_flags&RTF_REJECT) 2000 if (rt->rt6i_flags&RTF_REJECT)
1750 rtm->rtm_type = RTN_UNREACHABLE; 2001 rtm->rtm_type = RTN_UNREACHABLE;
1751 else if (rt->rt6i_dev && (rt->rt6i_dev->flags&IFF_LOOPBACK)) 2002 else if (rt->rt6i_dev && (rt->rt6i_dev->flags&IFF_LOOPBACK))
@@ -1766,31 +2017,35 @@ static int rt6_fill_node(struct sk_buff *skb, struct rt6_info *rt,
1766 rtm->rtm_flags |= RTM_F_CLONED; 2017 rtm->rtm_flags |= RTM_F_CLONED;
1767 2018
1768 if (dst) { 2019 if (dst) {
1769 RTA_PUT(skb, RTA_DST, 16, dst); 2020 NLA_PUT(skb, RTA_DST, 16, dst);
1770 rtm->rtm_dst_len = 128; 2021 rtm->rtm_dst_len = 128;
1771 } else if (rtm->rtm_dst_len) 2022 } else if (rtm->rtm_dst_len)
1772 RTA_PUT(skb, RTA_DST, 16, &rt->rt6i_dst.addr); 2023 NLA_PUT(skb, RTA_DST, 16, &rt->rt6i_dst.addr);
1773#ifdef CONFIG_IPV6_SUBTREES 2024#ifdef CONFIG_IPV6_SUBTREES
1774 if (src) { 2025 if (src) {
1775 RTA_PUT(skb, RTA_SRC, 16, src); 2026 NLA_PUT(skb, RTA_SRC, 16, src);
1776 rtm->rtm_src_len = 128; 2027 rtm->rtm_src_len = 128;
1777 } else if (rtm->rtm_src_len) 2028 } else if (rtm->rtm_src_len)
1778 RTA_PUT(skb, RTA_SRC, 16, &rt->rt6i_src.addr); 2029 NLA_PUT(skb, RTA_SRC, 16, &rt->rt6i_src.addr);
1779#endif 2030#endif
1780 if (iif) 2031 if (iif)
1781 RTA_PUT(skb, RTA_IIF, 4, &iif); 2032 NLA_PUT_U32(skb, RTA_IIF, iif);
1782 else if (dst) { 2033 else if (dst) {
1783 struct in6_addr saddr_buf; 2034 struct in6_addr saddr_buf;
1784 if (ipv6_get_saddr(&rt->u.dst, dst, &saddr_buf) == 0) 2035 if (ipv6_get_saddr(&rt->u.dst, dst, &saddr_buf) == 0)
1785 RTA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf); 2036 NLA_PUT(skb, RTA_PREFSRC, 16, &saddr_buf);
1786 } 2037 }
2038
1787 if (rtnetlink_put_metrics(skb, rt->u.dst.metrics) < 0) 2039 if (rtnetlink_put_metrics(skb, rt->u.dst.metrics) < 0)
1788 goto rtattr_failure; 2040 goto nla_put_failure;
2041
1789 if (rt->u.dst.neighbour) 2042 if (rt->u.dst.neighbour)
1790 RTA_PUT(skb, RTA_GATEWAY, 16, &rt->u.dst.neighbour->primary_key); 2043 NLA_PUT(skb, RTA_GATEWAY, 16, &rt->u.dst.neighbour->primary_key);
2044
1791 if (rt->u.dst.dev) 2045 if (rt->u.dst.dev)
1792 RTA_PUT(skb, RTA_OIF, sizeof(int), &rt->rt6i_dev->ifindex); 2046 NLA_PUT_U32(skb, RTA_OIF, rt->rt6i_dev->ifindex);
1793 RTA_PUT(skb, RTA_PRIORITY, 4, &rt->rt6i_metric); 2047
2048 NLA_PUT_U32(skb, RTA_PRIORITY, rt->rt6i_metric);
1794 ci.rta_lastuse = jiffies_to_clock_t(jiffies - rt->u.dst.lastuse); 2049 ci.rta_lastuse = jiffies_to_clock_t(jiffies - rt->u.dst.lastuse);
1795 if (rt->rt6i_expires) 2050 if (rt->rt6i_expires)
1796 ci.rta_expires = jiffies_to_clock_t(rt->rt6i_expires - jiffies); 2051 ci.rta_expires = jiffies_to_clock_t(rt->rt6i_expires - jiffies);
@@ -1802,23 +2057,21 @@ static int rt6_fill_node(struct sk_buff *skb, struct rt6_info *rt,
1802 ci.rta_id = 0; 2057 ci.rta_id = 0;
1803 ci.rta_ts = 0; 2058 ci.rta_ts = 0;
1804 ci.rta_tsage = 0; 2059 ci.rta_tsage = 0;
1805 RTA_PUT(skb, RTA_CACHEINFO, sizeof(ci), &ci); 2060 NLA_PUT(skb, RTA_CACHEINFO, sizeof(ci), &ci);
1806 nlh->nlmsg_len = skb->tail - b;
1807 return skb->len;
1808 2061
1809nlmsg_failure: 2062 return nlmsg_end(skb, nlh);
1810rtattr_failure: 2063
1811 skb_trim(skb, b - skb->data); 2064nla_put_failure:
1812 return -1; 2065 return nlmsg_cancel(skb, nlh);
1813} 2066}
1814 2067
1815static int rt6_dump_route(struct rt6_info *rt, void *p_arg) 2068int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1816{ 2069{
1817 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg; 2070 struct rt6_rtnl_dump_arg *arg = (struct rt6_rtnl_dump_arg *) p_arg;
1818 int prefix; 2071 int prefix;
1819 2072
1820 if (arg->cb->nlh->nlmsg_len >= NLMSG_LENGTH(sizeof(struct rtmsg))) { 2073 if (nlmsg_len(arg->cb->nlh) >= sizeof(struct rtmsg)) {
1821 struct rtmsg *rtm = NLMSG_DATA(arg->cb->nlh); 2074 struct rtmsg *rtm = nlmsg_data(arg->cb->nlh);
1822 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0; 2075 prefix = (rtm->rtm_flags & RTM_F_PREFIX) != 0;
1823 } else 2076 } else
1824 prefix = 0; 2077 prefix = 0;
@@ -1828,189 +2081,108 @@ static int rt6_dump_route(struct rt6_info *rt, void *p_arg)
1828 prefix, NLM_F_MULTI); 2081 prefix, NLM_F_MULTI);
1829} 2082}
1830 2083
1831static int fib6_dump_node(struct fib6_walker_t *w) 2084int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg)
1832{ 2085{
1833 int res; 2086 struct nlattr *tb[RTA_MAX+1];
1834 struct rt6_info *rt; 2087 struct rt6_info *rt;
2088 struct sk_buff *skb;
2089 struct rtmsg *rtm;
2090 struct flowi fl;
2091 int err, iif = 0;
1835 2092
1836 for (rt = w->leaf; rt; rt = rt->u.next) { 2093 err = nlmsg_parse(nlh, sizeof(*rtm), tb, RTA_MAX, rtm_ipv6_policy);
1837 res = rt6_dump_route(rt, w->args); 2094 if (err < 0)
1838 if (res < 0) { 2095 goto errout;
1839 /* Frame is full, suspend walking */
1840 w->leaf = rt;
1841 return 1;
1842 }
1843 BUG_TRAP(res!=0);
1844 }
1845 w->leaf = NULL;
1846 return 0;
1847}
1848
1849static void fib6_dump_end(struct netlink_callback *cb)
1850{
1851 struct fib6_walker_t *w = (void*)cb->args[0];
1852
1853 if (w) {
1854 cb->args[0] = 0;
1855 fib6_walker_unlink(w);
1856 kfree(w);
1857 }
1858 cb->done = (void*)cb->args[1];
1859 cb->args[1] = 0;
1860}
1861
1862static int fib6_dump_done(struct netlink_callback *cb)
1863{
1864 fib6_dump_end(cb);
1865 return cb->done ? cb->done(cb) : 0;
1866}
1867
1868int inet6_dump_fib(struct sk_buff *skb, struct netlink_callback *cb)
1869{
1870 struct rt6_rtnl_dump_arg arg;
1871 struct fib6_walker_t *w;
1872 int res;
1873 2096
1874 arg.skb = skb; 2097 err = -EINVAL;
1875 arg.cb = cb; 2098 memset(&fl, 0, sizeof(fl));
1876 2099
1877 w = (void*)cb->args[0]; 2100 if (tb[RTA_SRC]) {
1878 if (w == NULL) { 2101 if (nla_len(tb[RTA_SRC]) < sizeof(struct in6_addr))
1879 /* New dump: 2102 goto errout;
1880 *
1881 * 1. hook callback destructor.
1882 */
1883 cb->args[1] = (long)cb->done;
1884 cb->done = fib6_dump_done;
1885 2103
1886 /* 2104 ipv6_addr_copy(&fl.fl6_src, nla_data(tb[RTA_SRC]));
1887 * 2. allocate and initialize walker.
1888 */
1889 w = kzalloc(sizeof(*w), GFP_ATOMIC);
1890 if (w == NULL)
1891 return -ENOMEM;
1892 RT6_TRACE("dump<%p", w);
1893 w->root = &ip6_routing_table;
1894 w->func = fib6_dump_node;
1895 w->args = &arg;
1896 cb->args[0] = (long)w;
1897 read_lock_bh(&rt6_lock);
1898 res = fib6_walk(w);
1899 read_unlock_bh(&rt6_lock);
1900 } else {
1901 w->args = &arg;
1902 read_lock_bh(&rt6_lock);
1903 res = fib6_walk_continue(w);
1904 read_unlock_bh(&rt6_lock);
1905 } 2105 }
1906#if RT6_DEBUG >= 3
1907 if (res <= 0 && skb->len == 0)
1908 RT6_TRACE("%p>dump end\n", w);
1909#endif
1910 res = res < 0 ? res : skb->len;
1911 /* res < 0 is an error. (really, impossible)
1912 res == 0 means that dump is complete, but skb still can contain data.
1913 res > 0 dump is not complete, but frame is full.
1914 */
1915 /* Destroy walker, if dump of this table is complete. */
1916 if (res <= 0)
1917 fib6_dump_end(cb);
1918 return res;
1919}
1920
1921int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr* nlh, void *arg)
1922{
1923 struct rtattr **rta = arg;
1924 int iif = 0;
1925 int err = -ENOBUFS;
1926 struct sk_buff *skb;
1927 struct flowi fl;
1928 struct rt6_info *rt;
1929 2106
1930 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); 2107 if (tb[RTA_DST]) {
1931 if (skb == NULL) 2108 if (nla_len(tb[RTA_DST]) < sizeof(struct in6_addr))
1932 goto out; 2109 goto errout;
1933 2110
1934 /* Reserve room for dummy headers, this skb can pass 2111 ipv6_addr_copy(&fl.fl6_dst, nla_data(tb[RTA_DST]));
1935 through good chunk of routing engine. 2112 }
1936 */
1937 skb->mac.raw = skb->data;
1938 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1939 2113
1940 memset(&fl, 0, sizeof(fl)); 2114 if (tb[RTA_IIF])
1941 if (rta[RTA_SRC-1]) 2115 iif = nla_get_u32(tb[RTA_IIF]);
1942 ipv6_addr_copy(&fl.fl6_src,
1943 (struct in6_addr*)RTA_DATA(rta[RTA_SRC-1]));
1944 if (rta[RTA_DST-1])
1945 ipv6_addr_copy(&fl.fl6_dst,
1946 (struct in6_addr*)RTA_DATA(rta[RTA_DST-1]));
1947 2116
1948 if (rta[RTA_IIF-1]) 2117 if (tb[RTA_OIF])
1949 memcpy(&iif, RTA_DATA(rta[RTA_IIF-1]), sizeof(int)); 2118 fl.oif = nla_get_u32(tb[RTA_OIF]);
1950 2119
1951 if (iif) { 2120 if (iif) {
1952 struct net_device *dev; 2121 struct net_device *dev;
1953 dev = __dev_get_by_index(iif); 2122 dev = __dev_get_by_index(iif);
1954 if (!dev) { 2123 if (!dev) {
1955 err = -ENODEV; 2124 err = -ENODEV;
1956 goto out_free; 2125 goto errout;
1957 } 2126 }
1958 } 2127 }
1959 2128
1960 fl.oif = 0; 2129 skb = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL);
1961 if (rta[RTA_OIF-1]) 2130 if (skb == NULL) {
1962 memcpy(&fl.oif, RTA_DATA(rta[RTA_OIF-1]), sizeof(int)); 2131 err = -ENOBUFS;
2132 goto errout;
2133 }
1963 2134
1964 rt = (struct rt6_info*)ip6_route_output(NULL, &fl); 2135 /* Reserve room for dummy headers, this skb can pass
2136 through good chunk of routing engine.
2137 */
2138 skb->mac.raw = skb->data;
2139 skb_reserve(skb, MAX_HEADER + sizeof(struct ipv6hdr));
1965 2140
2141 rt = (struct rt6_info*) ip6_route_output(NULL, &fl);
1966 skb->dst = &rt->u.dst; 2142 skb->dst = &rt->u.dst;
1967 2143
1968 NETLINK_CB(skb).dst_pid = NETLINK_CB(in_skb).pid; 2144 err = rt6_fill_node(skb, rt, &fl.fl6_dst, &fl.fl6_src, iif,
1969 err = rt6_fill_node(skb, rt,
1970 &fl.fl6_dst, &fl.fl6_src,
1971 iif,
1972 RTM_NEWROUTE, NETLINK_CB(in_skb).pid, 2145 RTM_NEWROUTE, NETLINK_CB(in_skb).pid,
1973 nlh->nlmsg_seq, 0, 0); 2146 nlh->nlmsg_seq, 0, 0);
1974 if (err < 0) { 2147 if (err < 0) {
1975 err = -EMSGSIZE; 2148 kfree_skb(skb);
1976 goto out_free; 2149 goto errout;
1977 } 2150 }
1978 2151
1979 err = netlink_unicast(rtnl, skb, NETLINK_CB(in_skb).pid, MSG_DONTWAIT); 2152 err = rtnl_unicast(skb, NETLINK_CB(in_skb).pid);
1980 if (err > 0) 2153errout:
1981 err = 0;
1982out:
1983 return err; 2154 return err;
1984out_free:
1985 kfree_skb(skb);
1986 goto out;
1987} 2155}
1988 2156
1989void inet6_rt_notify(int event, struct rt6_info *rt, struct nlmsghdr *nlh, 2157void inet6_rt_notify(int event, struct rt6_info *rt, struct nl_info *info)
1990 struct netlink_skb_parms *req)
1991{ 2158{
1992 struct sk_buff *skb; 2159 struct sk_buff *skb;
1993 int size = NLMSG_SPACE(sizeof(struct rtmsg)+256); 2160 u32 pid = 0, seq = 0;
1994 u32 pid = current->pid; 2161 struct nlmsghdr *nlh = NULL;
1995 u32 seq = 0; 2162 int payload = sizeof(struct rtmsg) + 256;
1996 2163 int err = -ENOBUFS;
1997 if (req) 2164
1998 pid = req->pid; 2165 if (info) {
1999 if (nlh) 2166 pid = info->pid;
2000 seq = nlh->nlmsg_seq; 2167 nlh = info->nlh;
2001 2168 if (nlh)
2002 skb = alloc_skb(size, gfp_any()); 2169 seq = nlh->nlmsg_seq;
2003 if (!skb) {
2004 netlink_set_err(rtnl, 0, RTNLGRP_IPV6_ROUTE, ENOBUFS);
2005 return;
2006 } 2170 }
2007 if (rt6_fill_node(skb, rt, NULL, NULL, 0, event, pid, seq, 0, 0) < 0) { 2171
2172 skb = nlmsg_new(nlmsg_total_size(payload), gfp_any());
2173 if (skb == NULL)
2174 goto errout;
2175
2176 err = rt6_fill_node(skb, rt, NULL, NULL, 0, event, pid, seq, 0, 0);
2177 if (err < 0) {
2008 kfree_skb(skb); 2178 kfree_skb(skb);
2009 netlink_set_err(rtnl, 0, RTNLGRP_IPV6_ROUTE, EINVAL); 2179 goto errout;
2010 return;
2011 } 2180 }
2012 NETLINK_CB(skb).dst_group = RTNLGRP_IPV6_ROUTE; 2181
2013 netlink_broadcast(rtnl, skb, 0, RTNLGRP_IPV6_ROUTE, gfp_any()); 2182 err = rtnl_notify(skb, pid, RTNLGRP_IPV6_ROUTE, nlh, gfp_any());
2183errout:
2184 if (err < 0)
2185 rtnl_set_sk_err(RTNLGRP_IPV6_ROUTE, err);
2014} 2186}
2015 2187
2016/* 2188/*
@@ -2086,16 +2258,13 @@ static int rt6_info_route(struct rt6_info *rt, void *p_arg)
2086 2258
2087static int rt6_proc_info(char *buffer, char **start, off_t offset, int length) 2259static int rt6_proc_info(char *buffer, char **start, off_t offset, int length)
2088{ 2260{
2089 struct rt6_proc_arg arg; 2261 struct rt6_proc_arg arg = {
2090 arg.buffer = buffer; 2262 .buffer = buffer,
2091 arg.offset = offset; 2263 .offset = offset,
2092 arg.length = length; 2264 .length = length,
2093 arg.skip = 0; 2265 };
2094 arg.len = 0;
2095 2266
2096 read_lock_bh(&rt6_lock); 2267 fib6_clean_all(rt6_info_route, 0, &arg);
2097 fib6_clean_tree(&ip6_routing_table, rt6_info_route, 0, &arg);
2098 read_unlock_bh(&rt6_lock);
2099 2268
2100 *start = buffer; 2269 *start = buffer;
2101 if (offset) 2270 if (offset)
@@ -2250,13 +2419,9 @@ void __init ip6_route_init(void)
2250{ 2419{
2251 struct proc_dir_entry *p; 2420 struct proc_dir_entry *p;
2252 2421
2253 ip6_dst_ops.kmem_cachep = kmem_cache_create("ip6_dst_cache", 2422 ip6_dst_ops.kmem_cachep =
2254 sizeof(struct rt6_info), 2423 kmem_cache_create("ip6_dst_cache", sizeof(struct rt6_info), 0,
2255 0, SLAB_HWCACHE_ALIGN, 2424 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL, NULL);
2256 NULL, NULL);
2257 if (!ip6_dst_ops.kmem_cachep)
2258 panic("cannot create ip6_dst_cache");
2259
2260 fib6_init(); 2425 fib6_init();
2261#ifdef CONFIG_PROC_FS 2426#ifdef CONFIG_PROC_FS
2262 p = proc_net_create("ipv6_route", 0, rt6_proc_info); 2427 p = proc_net_create("ipv6_route", 0, rt6_proc_info);
@@ -2268,10 +2433,16 @@ void __init ip6_route_init(void)
2268#ifdef CONFIG_XFRM 2433#ifdef CONFIG_XFRM
2269 xfrm6_init(); 2434 xfrm6_init();
2270#endif 2435#endif
2436#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2437 fib6_rules_init();
2438#endif
2271} 2439}
2272 2440
2273void ip6_route_cleanup(void) 2441void ip6_route_cleanup(void)
2274{ 2442{
2443#ifdef CONFIG_IPV6_MULTIPLE_TABLES
2444 fib6_rules_cleanup();
2445#endif
2275#ifdef CONFIG_PROC_FS 2446#ifdef CONFIG_PROC_FS
2276 proc_net_remove("ipv6_route"); 2447 proc_net_remove("ipv6_route");
2277 proc_net_remove("rt6_stats"); 2448 proc_net_remove("rt6_stats");
generated by cgit v1.2.2 (git 2.25.0) at 2025-12-04 11:32:04 -0500