route.c (f43dc23d5ea91fca257be02138a255f02d98e806) route.c (c7066f70d9610df0b9406cc635fc09e86136e714)
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 *
8 * Authors: Ross Biro

--- 125 unchanged lines hidden (view full) ---

134static struct delayed_work expires_work;
135static unsigned long expires_ljiffies;
136
137/*
138 * Interface to generic destination cache.
139 */
140
141static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie);
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 *
8 * Authors: Ross Biro

--- 125 unchanged lines hidden (view full) ---

134static struct delayed_work expires_work;
135static unsigned long expires_ljiffies;
136
137/*
138 * Interface to generic destination cache.
139 */
140
141static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie);
142static unsigned int ipv4_default_advmss(const struct dst_entry *dst);
143static unsigned int ipv4_default_mtu(const struct dst_entry *dst);
144static void ipv4_dst_destroy(struct dst_entry *dst);
145static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst);
146static void ipv4_link_failure(struct sk_buff *skb);
147static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu);
148static int rt_garbage_collect(struct dst_ops *ops);
149
150static void ipv4_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
151 int how)
152{
153}
154
155static struct dst_ops ipv4_dst_ops = {
156 .family = AF_INET,
157 .protocol = cpu_to_be16(ETH_P_IP),
158 .gc = rt_garbage_collect,
159 .check = ipv4_dst_check,
142static void ipv4_dst_destroy(struct dst_entry *dst);
143static struct dst_entry *ipv4_negative_advice(struct dst_entry *dst);
144static void ipv4_link_failure(struct sk_buff *skb);
145static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu);
146static int rt_garbage_collect(struct dst_ops *ops);
147
148static void ipv4_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
149 int how)
150{
151}
152
153static struct dst_ops ipv4_dst_ops = {
154 .family = AF_INET,
155 .protocol = cpu_to_be16(ETH_P_IP),
156 .gc = rt_garbage_collect,
157 .check = ipv4_dst_check,
160 .default_advmss = ipv4_default_advmss,
161 .default_mtu = ipv4_default_mtu,
162 .destroy = ipv4_dst_destroy,
163 .ifdown = ipv4_dst_ifdown,
164 .negative_advice = ipv4_negative_advice,
165 .link_failure = ipv4_link_failure,
166 .update_pmtu = ip_rt_update_pmtu,
167 .local_out = __ip_local_out,
168};
169

--- 212 unchanged lines hidden (view full) ---

382
383 seq_printf(seq, "%s\t%08X\t%08X\t%8X\t%d\t%u\t%d\t"
384 "%08X\t%d\t%u\t%u\t%02X\t%d\t%1d\t%08X%n",
385 r->dst.dev ? r->dst.dev->name : "*",
386 (__force u32)r->rt_dst,
387 (__force u32)r->rt_gateway,
388 r->rt_flags, atomic_read(&r->dst.__refcnt),
389 r->dst.__use, 0, (__force u32)r->rt_src,
158 .destroy = ipv4_dst_destroy,
159 .ifdown = ipv4_dst_ifdown,
160 .negative_advice = ipv4_negative_advice,
161 .link_failure = ipv4_link_failure,
162 .update_pmtu = ip_rt_update_pmtu,
163 .local_out = __ip_local_out,
164};
165

--- 212 unchanged lines hidden (view full) ---

378
379 seq_printf(seq, "%s\t%08X\t%08X\t%8X\t%d\t%u\t%d\t"
380 "%08X\t%d\t%u\t%u\t%02X\t%d\t%1d\t%08X%n",
381 r->dst.dev ? r->dst.dev->name : "*",
382 (__force u32)r->rt_dst,
383 (__force u32)r->rt_gateway,
384 r->rt_flags, atomic_read(&r->dst.__refcnt),
385 r->dst.__use, 0, (__force u32)r->rt_src,
390 dst_metric_advmss(&r->dst) + 40,
386 (dst_metric(&r->dst, RTAX_ADVMSS) ?
387 (int)dst_metric(&r->dst, RTAX_ADVMSS) + 40 : 0),
391 dst_metric(&r->dst, RTAX_WINDOW),
392 (int)((dst_metric(&r->dst, RTAX_RTT) >> 3) +
393 dst_metric(&r->dst, RTAX_RTTVAR)),
394 r->fl.fl4_tos,
395 r->dst.hh ? atomic_read(&r->dst.hh->hh_refcnt) : -1,
396 r->dst.hh ? (r->dst.hh->hh_output ==
397 dev_queue_xmit) : 0,
398 r->rt_spec_dst, &len);

--- 110 unchanged lines hidden (view full) ---

509static const struct file_operations rt_cpu_seq_fops = {
510 .owner = THIS_MODULE,
511 .open = rt_cpu_seq_open,
512 .read = seq_read,
513 .llseek = seq_lseek,
514 .release = seq_release,
515};
516
388 dst_metric(&r->dst, RTAX_WINDOW),
389 (int)((dst_metric(&r->dst, RTAX_RTT) >> 3) +
390 dst_metric(&r->dst, RTAX_RTTVAR)),
391 r->fl.fl4_tos,
392 r->dst.hh ? atomic_read(&r->dst.hh->hh_refcnt) : -1,
393 r->dst.hh ? (r->dst.hh->hh_output ==
394 dev_queue_xmit) : 0,
395 r->rt_spec_dst, &len);

--- 110 unchanged lines hidden (view full) ---

506static const struct file_operations rt_cpu_seq_fops = {
507 .owner = THIS_MODULE,
508 .open = rt_cpu_seq_open,
509 .read = seq_read,
510 .llseek = seq_lseek,
511 .release = seq_release,
512};
513
517#ifdef CONFIG_NET_CLS_ROUTE
514#ifdef CONFIG_IP_ROUTE_CLASSID
518static int rt_acct_proc_show(struct seq_file *m, void *v)
519{
520 struct ip_rt_acct *dst, *src;
521 unsigned int i, j;
522
523 dst = kcalloc(256, sizeof(struct ip_rt_acct), GFP_KERNEL);
524 if (!dst)
525 return -ENOMEM;

--- 36 unchanged lines hidden (view full) ---

562 if (!pde)
563 goto err1;
564
565 pde = proc_create("rt_cache", S_IRUGO,
566 net->proc_net_stat, &rt_cpu_seq_fops);
567 if (!pde)
568 goto err2;
569
515static int rt_acct_proc_show(struct seq_file *m, void *v)
516{
517 struct ip_rt_acct *dst, *src;
518 unsigned int i, j;
519
520 dst = kcalloc(256, sizeof(struct ip_rt_acct), GFP_KERNEL);
521 if (!dst)
522 return -ENOMEM;

--- 36 unchanged lines hidden (view full) ---

559 if (!pde)
560 goto err1;
561
562 pde = proc_create("rt_cache", S_IRUGO,
563 net->proc_net_stat, &rt_cpu_seq_fops);
564 if (!pde)
565 goto err2;
566
570#ifdef CONFIG_NET_CLS_ROUTE
567#ifdef CONFIG_IP_ROUTE_CLASSID
571 pde = proc_create("rt_acct", 0, net->proc_net, &rt_acct_proc_fops);
572 if (!pde)
573 goto err3;
574#endif
575 return 0;
576
568 pde = proc_create("rt_acct", 0, net->proc_net, &rt_acct_proc_fops);
569 if (!pde)
570 goto err3;
571#endif
572 return 0;
573
577#ifdef CONFIG_NET_CLS_ROUTE
574#ifdef CONFIG_IP_ROUTE_CLASSID
578err3:
579 remove_proc_entry("rt_cache", net->proc_net_stat);
580#endif
581err2:
582 remove_proc_entry("rt_cache", net->proc_net);
583err1:
584 return -ENOMEM;
585}
586
587static void __net_exit ip_rt_do_proc_exit(struct net *net)
588{
589 remove_proc_entry("rt_cache", net->proc_net_stat);
590 remove_proc_entry("rt_cache", net->proc_net);
575err3:
576 remove_proc_entry("rt_cache", net->proc_net_stat);
577#endif
578err2:
579 remove_proc_entry("rt_cache", net->proc_net);
580err1:
581 return -ENOMEM;
582}
583
584static void __net_exit ip_rt_do_proc_exit(struct net *net)
585{
586 remove_proc_entry("rt_cache", net->proc_net_stat);
587 remove_proc_entry("rt_cache", net->proc_net);
591#ifdef CONFIG_NET_CLS_ROUTE
588#ifdef CONFIG_IP_ROUTE_CLASSID
592 remove_proc_entry("rt_acct", net->proc_net);
593#endif
594}
595
596static struct pernet_operations ip_rt_proc_ops __net_initdata = {
597 .init = ip_rt_do_proc_init,
598 .exit = ip_rt_do_proc_exit,
599};

--- 82 unchanged lines hidden (view full) ---

682{
683 return net->ipv4.current_rt_cache_rebuild_count <=
684 net->ipv4.sysctl_rt_cache_rebuild_count;
685}
686
687static inline bool compare_hash_inputs(const struct flowi *fl1,
688 const struct flowi *fl2)
689{
589 remove_proc_entry("rt_acct", net->proc_net);
590#endif
591}
592
593static struct pernet_operations ip_rt_proc_ops __net_initdata = {
594 .init = ip_rt_do_proc_init,
595 .exit = ip_rt_do_proc_exit,
596};

--- 82 unchanged lines hidden (view full) ---

679{
680 return net->ipv4.current_rt_cache_rebuild_count <=
681 net->ipv4.sysctl_rt_cache_rebuild_count;
682}
683
684static inline bool compare_hash_inputs(const struct flowi *fl1,
685 const struct flowi *fl2)
686{
690 return ((((__force u32)fl1->fl4_dst ^ (__force u32)fl2->fl4_dst) |
691 ((__force u32)fl1->fl4_src ^ (__force u32)fl2->fl4_src) |
687 return ((((__force u32)fl1->nl_u.ip4_u.daddr ^ (__force u32)fl2->nl_u.ip4_u.daddr) |
688 ((__force u32)fl1->nl_u.ip4_u.saddr ^ (__force u32)fl2->nl_u.ip4_u.saddr) |
692 (fl1->iif ^ fl2->iif)) == 0);
693}
694
695static inline int compare_keys(struct flowi *fl1, struct flowi *fl2)
696{
689 (fl1->iif ^ fl2->iif)) == 0);
690}
691
692static inline int compare_keys(struct flowi *fl1, struct flowi *fl2)
693{
697 return (((__force u32)fl1->fl4_dst ^ (__force u32)fl2->fl4_dst) |
698 ((__force u32)fl1->fl4_src ^ (__force u32)fl2->fl4_src) |
694 return (((__force u32)fl1->nl_u.ip4_u.daddr ^ (__force u32)fl2->nl_u.ip4_u.daddr) |
695 ((__force u32)fl1->nl_u.ip4_u.saddr ^ (__force u32)fl2->nl_u.ip4_u.saddr) |
699 (fl1->mark ^ fl2->mark) |
696 (fl1->mark ^ fl2->mark) |
700 (*(u16 *)&fl1->fl4_tos ^ *(u16 *)&fl2->fl4_tos) |
697 (*(u16 *)&fl1->nl_u.ip4_u.tos ^ *(u16 *)&fl2->nl_u.ip4_u.tos) |
701 (fl1->oif ^ fl2->oif) |
702 (fl1->iif ^ fl2->iif)) == 0;
703}
704
705static inline int compare_netns(struct rtable *rt1, struct rtable *rt2)
706{
707 return net_eq(dev_net(rt1->dst.dev), dev_net(rt2->dst.dev));
708}
709
710static inline int rt_is_expired(struct rtable *rth)
711{
712 return rth->rt_genid != rt_genid(dev_net(rth->dst.dev));
713}
714
715/*
716 * Perform a full scan of hash table and free all entries.
717 * Can be called by a softirq or a process.
718 * In the later case, we want to be reschedule if necessary
719 */
698 (fl1->oif ^ fl2->oif) |
699 (fl1->iif ^ fl2->iif)) == 0;
700}
701
702static inline int compare_netns(struct rtable *rt1, struct rtable *rt2)
703{
704 return net_eq(dev_net(rt1->dst.dev), dev_net(rt2->dst.dev));
705}
706
707static inline int rt_is_expired(struct rtable *rth)
708{
709 return rth->rt_genid != rt_genid(dev_net(rth->dst.dev));
710}
711
712/*
713 * Perform a full scan of hash table and free all entries.
714 * Can be called by a softirq or a process.
715 * In the later case, we want to be reschedule if necessary
716 */
720static void rt_do_flush(struct net *net, int process_context)
717static void rt_do_flush(int process_context)
721{
722 unsigned int i;
723 struct rtable *rth, *next;
718{
719 unsigned int i;
720 struct rtable *rth, *next;
721 struct rtable * tail;
724
725 for (i = 0; i <= rt_hash_mask; i++) {
722
723 for (i = 0; i <= rt_hash_mask; i++) {
726 struct rtable __rcu **pprev;
727 struct rtable *list;
728
729 if (process_context && need_resched())
730 cond_resched();
731 rth = rcu_dereference_raw(rt_hash_table[i].chain);
732 if (!rth)
733 continue;
734
735 spin_lock_bh(rt_hash_lock_addr(i));
724 if (process_context && need_resched())
725 cond_resched();
726 rth = rcu_dereference_raw(rt_hash_table[i].chain);
727 if (!rth)
728 continue;
729
730 spin_lock_bh(rt_hash_lock_addr(i));
731#ifdef CONFIG_NET_NS
732 {
733 struct rtable __rcu **prev;
734 struct rtable *p;
736
735
737 list = NULL;
738 pprev = &rt_hash_table[i].chain;
739 rth = rcu_dereference_protected(*pprev,
736 rth = rcu_dereference_protected(rt_hash_table[i].chain,
740 lockdep_is_held(rt_hash_lock_addr(i)));
741
737 lockdep_is_held(rt_hash_lock_addr(i)));
738
742 while (rth) {
743 next = rcu_dereference_protected(rth->dst.rt_next,
744 lockdep_is_held(rt_hash_lock_addr(i)));
739 /* defer releasing the head of the list after spin_unlock */
740 for (tail = rth; tail;
741 tail = rcu_dereference_protected(tail->dst.rt_next,
742 lockdep_is_held(rt_hash_lock_addr(i))))
743 if (!rt_is_expired(tail))
744 break;
745 if (rth != tail)
746 rt_hash_table[i].chain = tail;
745
747
746 if (!net ||
747 net_eq(dev_net(rth->dst.dev), net)) {
748 rcu_assign_pointer(*pprev, next);
749 rcu_assign_pointer(rth->dst.rt_next, list);
750 list = rth;
748 /* call rt_free on entries after the tail requiring flush */
749 prev = &rt_hash_table[i].chain;
750 for (p = rcu_dereference_protected(*prev,
751 lockdep_is_held(rt_hash_lock_addr(i)));
752 p != NULL;
753 p = next) {
754 next = rcu_dereference_protected(p->dst.rt_next,
755 lockdep_is_held(rt_hash_lock_addr(i)));
756 if (!rt_is_expired(p)) {
757 prev = &p->dst.rt_next;
751 } else {
758 } else {
752 pprev = &rth->dst.rt_next;
759 *prev = next;
760 rt_free(p);
753 }
761 }
754 rth = next;
755 }
762 }
756
763 }
764#else
765 rth = rcu_dereference_protected(rt_hash_table[i].chain,
766 lockdep_is_held(rt_hash_lock_addr(i)));
767 rcu_assign_pointer(rt_hash_table[i].chain, NULL);
768 tail = NULL;
769#endif
757 spin_unlock_bh(rt_hash_lock_addr(i));
758
770 spin_unlock_bh(rt_hash_lock_addr(i));
771
759 for (; list; list = next) {
760 next = rcu_dereference_protected(list->dst.rt_next, 1);
761 rt_free(list);
772 for (; rth != tail; rth = next) {
773 next = rcu_dereference_protected(rth->dst.rt_next, 1);
774 rt_free(rth);
762 }
763 }
764}
765
766/*
767 * While freeing expired entries, we compute average chain length
768 * and standard deviation, using fixed-point arithmetic.
769 * This to have an estimation of rt_chain_length_max

--- 131 unchanged lines hidden (view full) ---

901/*
902 * delay < 0 : invalidate cache (fast : entries will be deleted later)
903 * delay >= 0 : invalidate & flush cache (can be long)
904 */
905void rt_cache_flush(struct net *net, int delay)
906{
907 rt_cache_invalidate(net);
908 if (delay >= 0)
775 }
776 }
777}
778
779/*
780 * While freeing expired entries, we compute average chain length
781 * and standard deviation, using fixed-point arithmetic.
782 * This to have an estimation of rt_chain_length_max

--- 131 unchanged lines hidden (view full) ---

914/*
915 * delay < 0 : invalidate cache (fast : entries will be deleted later)
916 * delay >= 0 : invalidate & flush cache (can be long)
917 */
918void rt_cache_flush(struct net *net, int delay)
919{
920 rt_cache_invalidate(net);
921 if (delay >= 0)
909 rt_do_flush(net, !in_softirq());
922 rt_do_flush(!in_softirq());
910}
911
912/* Flush previous cache invalidated entries from the cache */
923}
924
925/* Flush previous cache invalidated entries from the cache */
913void rt_cache_flush_batch(struct net *net)
926void rt_cache_flush_batch(void)
914{
927{
915 rt_do_flush(net, !in_softirq());
928 rt_do_flush(!in_softirq());
916}
917
918static void rt_emergency_hash_rebuild(struct net *net)
919{
920 if (net_ratelimit())
921 printk(KERN_WARNING "Route hash chain too long!\n");
922 rt_cache_invalidate(net);
923}

--- 347 unchanged lines hidden (view full) ---

1271 skb_dst_set(skb, &rt->dst);
1272 return 0;
1273}
1274
1275void rt_bind_peer(struct rtable *rt, int create)
1276{
1277 struct inet_peer *peer;
1278
929}
930
931static void rt_emergency_hash_rebuild(struct net *net)
932{
933 if (net_ratelimit())
934 printk(KERN_WARNING "Route hash chain too long!\n");
935 rt_cache_invalidate(net);
936}

--- 347 unchanged lines hidden (view full) ---

1284 skb_dst_set(skb, &rt->dst);
1285 return 0;
1286}
1287
1288void rt_bind_peer(struct rtable *rt, int create)
1289{
1290 struct inet_peer *peer;
1291
1279 peer = inet_getpeer_v4(rt->rt_dst, create);
1292 peer = inet_getpeer(rt->rt_dst, create);
1280
1281 if (peer && cmpxchg(&rt->peer, NULL, peer) != NULL)
1282 inet_putpeer(peer);
1283}
1284
1285/*
1286 * Peer allocation may fail only in serious out-of-memory conditions. However
1287 * we still can generate some output.

--- 380 unchanged lines hidden (view full) ---

1668 old_mtu -= iph->ihl << 2;
1669
1670 mtu = guess_mtu(old_mtu);
1671 }
1672 if (mtu <= dst_mtu(&rth->dst)) {
1673 if (mtu < dst_mtu(&rth->dst)) {
1674 dst_confirm(&rth->dst);
1675 if (mtu < ip_rt_min_pmtu) {
1293
1294 if (peer && cmpxchg(&rt->peer, NULL, peer) != NULL)
1295 inet_putpeer(peer);
1296}
1297
1298/*
1299 * Peer allocation may fail only in serious out-of-memory conditions. However
1300 * we still can generate some output.

--- 380 unchanged lines hidden (view full) ---

1681 old_mtu -= iph->ihl << 2;
1682
1683 mtu = guess_mtu(old_mtu);
1684 }
1685 if (mtu <= dst_mtu(&rth->dst)) {
1686 if (mtu < dst_mtu(&rth->dst)) {
1687 dst_confirm(&rth->dst);
1688 if (mtu < ip_rt_min_pmtu) {
1676 u32 lock = dst_metric(&rth->dst,
1677 RTAX_LOCK);
1678 mtu = ip_rt_min_pmtu;
1689 mtu = ip_rt_min_pmtu;
1679 lock |= (1 << RTAX_MTU);
1680 dst_metric_set(&rth->dst, RTAX_LOCK,
1681 lock);
1690 rth->dst.metrics[RTAX_LOCK-1] |=
1691 (1 << RTAX_MTU);
1682 }
1692 }
1683 dst_metric_set(&rth->dst, RTAX_MTU, mtu);
1693 rth->dst.metrics[RTAX_MTU-1] = mtu;
1684 dst_set_expires(&rth->dst,
1685 ip_rt_mtu_expires);
1686 }
1687 est_mtu = mtu;
1688 }
1689 }
1690 rcu_read_unlock();
1691 }
1692 }
1693 return est_mtu ? : new_mtu;
1694}
1695
1696static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
1697{
1698 if (dst_mtu(dst) > mtu && mtu >= 68 &&
1699 !(dst_metric_locked(dst, RTAX_MTU))) {
1700 if (mtu < ip_rt_min_pmtu) {
1694 dst_set_expires(&rth->dst,
1695 ip_rt_mtu_expires);
1696 }
1697 est_mtu = mtu;
1698 }
1699 }
1700 rcu_read_unlock();
1701 }
1702 }
1703 return est_mtu ? : new_mtu;
1704}
1705
1706static void ip_rt_update_pmtu(struct dst_entry *dst, u32 mtu)
1707{
1708 if (dst_mtu(dst) > mtu && mtu >= 68 &&
1709 !(dst_metric_locked(dst, RTAX_MTU))) {
1710 if (mtu < ip_rt_min_pmtu) {
1701 u32 lock = dst_metric(dst, RTAX_LOCK);
1702 mtu = ip_rt_min_pmtu;
1711 mtu = ip_rt_min_pmtu;
1703 dst_metric_set(dst, RTAX_LOCK, lock | (1 << RTAX_MTU));
1712 dst->metrics[RTAX_LOCK-1] |= (1 << RTAX_MTU);
1704 }
1713 }
1705 dst_metric_set(dst, RTAX_MTU, mtu);
1714 dst->metrics[RTAX_MTU-1] = mtu;
1706 dst_set_expires(dst, ip_rt_mtu_expires);
1707 call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst);
1708 }
1709}
1710
1711static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie)
1712{
1713 if (rt_is_expired((struct rtable *)dst))

--- 56 unchanged lines hidden (view full) ---

1770 else
1771 src = inet_select_addr(rt->dst.dev, rt->rt_gateway,
1772 RT_SCOPE_UNIVERSE);
1773 rcu_read_unlock();
1774 }
1775 memcpy(addr, &src, 4);
1776}
1777
1715 dst_set_expires(dst, ip_rt_mtu_expires);
1716 call_netevent_notifiers(NETEVENT_PMTU_UPDATE, dst);
1717 }
1718}
1719
1720static struct dst_entry *ipv4_dst_check(struct dst_entry *dst, u32 cookie)
1721{
1722 if (rt_is_expired((struct rtable *)dst))

--- 56 unchanged lines hidden (view full) ---

1779 else
1780 src = inet_select_addr(rt->dst.dev, rt->rt_gateway,
1781 RT_SCOPE_UNIVERSE);
1782 rcu_read_unlock();
1783 }
1784 memcpy(addr, &src, 4);
1785}
1786
1778#ifdef CONFIG_NET_CLS_ROUTE
1787#ifdef CONFIG_IP_ROUTE_CLASSID
1779static void set_class_tag(struct rtable *rt, u32 tag)
1780{
1781 if (!(rt->dst.tclassid & 0xFFFF))
1782 rt->dst.tclassid |= tag & 0xFFFF;
1783 if (!(rt->dst.tclassid & 0xFFFF0000))
1784 rt->dst.tclassid |= tag & 0xFFFF0000;
1785}
1786#endif
1787
1788static void set_class_tag(struct rtable *rt, u32 tag)
1789{
1790 if (!(rt->dst.tclassid & 0xFFFF))
1791 rt->dst.tclassid |= tag & 0xFFFF;
1792 if (!(rt->dst.tclassid & 0xFFFF0000))
1793 rt->dst.tclassid |= tag & 0xFFFF0000;
1794}
1795#endif
1796
1788static unsigned int ipv4_default_advmss(const struct dst_entry *dst)
1789{
1790 unsigned int advmss = dst_metric_raw(dst, RTAX_ADVMSS);
1791
1792 if (advmss == 0) {
1793 advmss = max_t(unsigned int, dst->dev->mtu - 40,
1794 ip_rt_min_advmss);
1795 if (advmss > 65535 - 40)
1796 advmss = 65535 - 40;
1797 }
1798 return advmss;
1799}
1800
1801static unsigned int ipv4_default_mtu(const struct dst_entry *dst)
1802{
1803 unsigned int mtu = dst->dev->mtu;
1804
1805 if (unlikely(dst_metric_locked(dst, RTAX_MTU))) {
1806 const struct rtable *rt = (const struct rtable *) dst;
1807
1808 if (rt->rt_gateway != rt->rt_dst && mtu > 576)
1809 mtu = 576;
1810 }
1811
1812 if (mtu > IP_MAX_MTU)
1813 mtu = IP_MAX_MTU;
1814
1815 return mtu;
1816}
1817
1818static void rt_set_nexthop(struct rtable *rt, struct fib_result *res, u32 itag)
1819{
1797static void rt_set_nexthop(struct rtable *rt, struct fib_result *res, u32 itag)
1798{
1820 struct dst_entry *dst = &rt->dst;
1821 struct fib_info *fi = res->fi;
1822
1823 if (fi) {
1824 if (FIB_RES_GW(*res) &&
1825 FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK)
1826 rt->rt_gateway = FIB_RES_GW(*res);
1799 struct fib_info *fi = res->fi;
1800
1801 if (fi) {
1802 if (FIB_RES_GW(*res) &&
1803 FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK)
1804 rt->rt_gateway = FIB_RES_GW(*res);
1827 dst_import_metrics(dst, fi->fib_metrics);
1828#ifdef CONFIG_NET_CLS_ROUTE
1829 dst->tclassid = FIB_RES_NH(*res).nh_tclassid;
1805 memcpy(rt->dst.metrics, fi->fib_metrics,
1806 sizeof(rt->dst.metrics));
1807 if (fi->fib_mtu == 0) {
1808 rt->dst.metrics[RTAX_MTU-1] = rt->dst.dev->mtu;
1809 if (dst_metric_locked(&rt->dst, RTAX_MTU) &&
1810 rt->rt_gateway != rt->rt_dst &&
1811 rt->dst.dev->mtu > 576)
1812 rt->dst.metrics[RTAX_MTU-1] = 576;
1813 }
1814#ifdef CONFIG_IP_ROUTE_CLASSID
1815 rt->dst.tclassid = FIB_RES_NH(*res).nh_tclassid;
1830#endif
1816#endif
1831 }
1817 } else
1818 rt->dst.metrics[RTAX_MTU-1]= rt->dst.dev->mtu;
1832
1819
1833 if (dst_mtu(dst) > IP_MAX_MTU)
1834 dst_metric_set(dst, RTAX_MTU, IP_MAX_MTU);
1835 if (dst_metric_raw(dst, RTAX_ADVMSS) > 65535 - 40)
1836 dst_metric_set(dst, RTAX_ADVMSS, 65535 - 40);
1820 if (dst_metric(&rt->dst, RTAX_HOPLIMIT) == 0)
1821 rt->dst.metrics[RTAX_HOPLIMIT-1] = sysctl_ip_default_ttl;
1822 if (dst_mtu(&rt->dst) > IP_MAX_MTU)
1823 rt->dst.metrics[RTAX_MTU-1] = IP_MAX_MTU;
1824 if (dst_metric(&rt->dst, RTAX_ADVMSS) == 0)
1825 rt->dst.metrics[RTAX_ADVMSS-1] = max_t(unsigned int, rt->dst.dev->mtu - 40,
1826 ip_rt_min_advmss);
1827 if (dst_metric(&rt->dst, RTAX_ADVMSS) > 65535 - 40)
1828 rt->dst.metrics[RTAX_ADVMSS-1] = 65535 - 40;
1837
1829
1838#ifdef CONFIG_NET_CLS_ROUTE
1830#ifdef CONFIG_IP_ROUTE_CLASSID
1839#ifdef CONFIG_IP_MULTIPLE_TABLES
1840 set_class_tag(rt, fib_rules_tclass(res));
1841#endif
1842 set_class_tag(rt, itag);
1843#endif
1844 rt->rt_type = res->type;
1845}
1846

--- 39 unchanged lines hidden (view full) ---

1886 if (IN_DEV_CONF_GET(in_dev, NOPOLICY))
1887 rth->dst.flags |= DST_NOPOLICY;
1888 rth->fl.fl4_dst = daddr;
1889 rth->rt_dst = daddr;
1890 rth->fl.fl4_tos = tos;
1891 rth->fl.mark = skb->mark;
1892 rth->fl.fl4_src = saddr;
1893 rth->rt_src = saddr;
1831#ifdef CONFIG_IP_MULTIPLE_TABLES
1832 set_class_tag(rt, fib_rules_tclass(res));
1833#endif
1834 set_class_tag(rt, itag);
1835#endif
1836 rt->rt_type = res->type;
1837}
1838

--- 39 unchanged lines hidden (view full) ---

1878 if (IN_DEV_CONF_GET(in_dev, NOPOLICY))
1879 rth->dst.flags |= DST_NOPOLICY;
1880 rth->fl.fl4_dst = daddr;
1881 rth->rt_dst = daddr;
1882 rth->fl.fl4_tos = tos;
1883 rth->fl.mark = skb->mark;
1884 rth->fl.fl4_src = saddr;
1885 rth->rt_src = saddr;
1894#ifdef CONFIG_NET_CLS_ROUTE
1886#ifdef CONFIG_IP_ROUTE_CLASSID
1895 rth->dst.tclassid = itag;
1896#endif
1897 rth->rt_iif =
1898 rth->fl.iif = dev->ifindex;
1899 rth->dst.dev = init_net.loopback_dev;
1900 dev_hold(rth->dst.dev);
1901 rth->fl.oif = 0;
1902 rth->rt_gateway = daddr;

--- 189 unchanged lines hidden (view full) ---

2092 * called with rcu_read_lock()
2093 */
2094
2095static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr,
2096 u8 tos, struct net_device *dev)
2097{
2098 struct fib_result res;
2099 struct in_device *in_dev = __in_dev_get_rcu(dev);
1887 rth->dst.tclassid = itag;
1888#endif
1889 rth->rt_iif =
1890 rth->fl.iif = dev->ifindex;
1891 rth->dst.dev = init_net.loopback_dev;
1892 dev_hold(rth->dst.dev);
1893 rth->fl.oif = 0;
1894 rth->rt_gateway = daddr;

--- 189 unchanged lines hidden (view full) ---

2084 * called with rcu_read_lock()
2085 */
2086
2087static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr,
2088 u8 tos, struct net_device *dev)
2089{
2090 struct fib_result res;
2091 struct in_device *in_dev = __in_dev_get_rcu(dev);
2100 struct flowi fl = { .fl4_dst = daddr,
2101 .fl4_src = saddr,
2102 .fl4_tos = tos,
2103 .fl4_scope = RT_SCOPE_UNIVERSE,
2092 struct flowi fl = { .nl_u = { .ip4_u =
2093 { .daddr = daddr,
2094 .saddr = saddr,
2095 .tos = tos,
2096 .scope = RT_SCOPE_UNIVERSE,
2097 } },
2104 .mark = skb->mark,
2105 .iif = dev->ifindex };
2106 unsigned flags = 0;
2107 u32 itag = 0;
2108 struct rtable * rth;
2109 unsigned hash;
2110 __be32 spec_dst;
2111 int err = -EINVAL;

--- 91 unchanged lines hidden (view full) ---

2203 if (IN_DEV_CONF_GET(in_dev, NOPOLICY))
2204 rth->dst.flags |= DST_NOPOLICY;
2205 rth->fl.fl4_dst = daddr;
2206 rth->rt_dst = daddr;
2207 rth->fl.fl4_tos = tos;
2208 rth->fl.mark = skb->mark;
2209 rth->fl.fl4_src = saddr;
2210 rth->rt_src = saddr;
2098 .mark = skb->mark,
2099 .iif = dev->ifindex };
2100 unsigned flags = 0;
2101 u32 itag = 0;
2102 struct rtable * rth;
2103 unsigned hash;
2104 __be32 spec_dst;
2105 int err = -EINVAL;

--- 91 unchanged lines hidden (view full) ---

2197 if (IN_DEV_CONF_GET(in_dev, NOPOLICY))
2198 rth->dst.flags |= DST_NOPOLICY;
2199 rth->fl.fl4_dst = daddr;
2200 rth->rt_dst = daddr;
2201 rth->fl.fl4_tos = tos;
2202 rth->fl.mark = skb->mark;
2203 rth->fl.fl4_src = saddr;
2204 rth->rt_src = saddr;
2211#ifdef CONFIG_NET_CLS_ROUTE
2205#ifdef CONFIG_IP_ROUTE_CLASSID
2212 rth->dst.tclassid = itag;
2213#endif
2214 rth->rt_iif =
2215 rth->fl.iif = dev->ifindex;
2216 rth->dst.dev = net->loopback_dev;
2217 dev_hold(rth->dst.dev);
2218 rth->rt_gateway = daddr;
2219 rth->rt_spec_dst= spec_dst;

--- 261 unchanged lines hidden (view full) ---

2481 * Major route resolver routine.
2482 * called with rcu_read_lock();
2483 */
2484
2485static int ip_route_output_slow(struct net *net, struct rtable **rp,
2486 const struct flowi *oldflp)
2487{
2488 u32 tos = RT_FL_TOS(oldflp);
2206 rth->dst.tclassid = itag;
2207#endif
2208 rth->rt_iif =
2209 rth->fl.iif = dev->ifindex;
2210 rth->dst.dev = net->loopback_dev;
2211 dev_hold(rth->dst.dev);
2212 rth->rt_gateway = daddr;
2213 rth->rt_spec_dst= spec_dst;

--- 261 unchanged lines hidden (view full) ---

2475 * Major route resolver routine.
2476 * called with rcu_read_lock();
2477 */
2478
2479static int ip_route_output_slow(struct net *net, struct rtable **rp,
2480 const struct flowi *oldflp)
2481{
2482 u32 tos = RT_FL_TOS(oldflp);
2489 struct flowi fl = { .fl4_dst = oldflp->fl4_dst,
2490 .fl4_src = oldflp->fl4_src,
2491 .fl4_tos = tos & IPTOS_RT_MASK,
2492 .fl4_scope = ((tos & RTO_ONLINK) ?
2493 RT_SCOPE_LINK : RT_SCOPE_UNIVERSE),
2483 struct flowi fl = { .nl_u = { .ip4_u =
2484 { .daddr = oldflp->fl4_dst,
2485 .saddr = oldflp->fl4_src,
2486 .tos = tos & IPTOS_RT_MASK,
2487 .scope = ((tos & RTO_ONLINK) ?
2488 RT_SCOPE_LINK :
2489 RT_SCOPE_UNIVERSE),
2490 } },
2494 .mark = oldflp->mark,
2495 .iif = net->loopback_dev->ifindex,
2496 .oif = oldflp->oif };
2497 struct fib_result res;
2498 unsigned int flags = 0;
2499 struct net_device *dev_out = NULL;
2500 int err;
2501

--- 55 unchanged lines hidden (view full) ---

2557
2558 if (oldflp->oif) {
2559 dev_out = dev_get_by_index_rcu(net, oldflp->oif);
2560 err = -ENODEV;
2561 if (dev_out == NULL)
2562 goto out;
2563
2564 /* RACE: Check return value of inet_select_addr instead. */
2491 .mark = oldflp->mark,
2492 .iif = net->loopback_dev->ifindex,
2493 .oif = oldflp->oif };
2494 struct fib_result res;
2495 unsigned int flags = 0;
2496 struct net_device *dev_out = NULL;
2497 int err;
2498

--- 55 unchanged lines hidden (view full) ---

2554
2555 if (oldflp->oif) {
2556 dev_out = dev_get_by_index_rcu(net, oldflp->oif);
2557 err = -ENODEV;
2558 if (dev_out == NULL)
2559 goto out;
2560
2561 /* RACE: Check return value of inet_select_addr instead. */
2565 if (!(dev_out->flags & IFF_UP) || !__in_dev_get_rcu(dev_out)) {
2566 err = -ENETUNREACH;
2567 goto out;
2568 }
2562 if (rcu_dereference(dev_out->ip_ptr) == NULL)
2563 goto out; /* Wrong error code */
2564
2569 if (ipv4_is_local_multicast(oldflp->fl4_dst) ||
2570 ipv4_is_lbcast(oldflp->fl4_dst)) {
2571 if (!fl.fl4_src)
2572 fl.fl4_src = inet_select_addr(dev_out, 0,
2573 RT_SCOPE_LINK);
2574 goto make_route;
2575 }
2576 if (!fl.fl4_src) {

--- 44 unchanged lines hidden (view full) ---

2621 res.type = RTN_UNICAST;
2622 goto make_route;
2623 }
2624 err = -ENETUNREACH;
2625 goto out;
2626 }
2627
2628 if (res.type == RTN_LOCAL) {
2565 if (ipv4_is_local_multicast(oldflp->fl4_dst) ||
2566 ipv4_is_lbcast(oldflp->fl4_dst)) {
2567 if (!fl.fl4_src)
2568 fl.fl4_src = inet_select_addr(dev_out, 0,
2569 RT_SCOPE_LINK);
2570 goto make_route;
2571 }
2572 if (!fl.fl4_src) {

--- 44 unchanged lines hidden (view full) ---

2617 res.type = RTN_UNICAST;
2618 goto make_route;
2619 }
2620 err = -ENETUNREACH;
2621 goto out;
2622 }
2623
2624 if (res.type == RTN_LOCAL) {
2629 if (!fl.fl4_src) {
2630 if (res.fi->fib_prefsrc)
2631 fl.fl4_src = res.fi->fib_prefsrc;
2632 else
2633 fl.fl4_src = fl.fl4_dst;
2634 }
2625 if (!fl.fl4_src)
2626 fl.fl4_src = fl.fl4_dst;
2635 dev_out = net->loopback_dev;
2636 fl.oif = dev_out->ifindex;
2637 res.fi = NULL;
2638 flags |= RTCF_LOCAL;
2639 goto make_route;
2640 }
2641
2642#ifdef CONFIG_IP_ROUTE_MULTIPATH

--- 85 unchanged lines hidden (view full) ---

2728
2729 if (rt) {
2730 struct dst_entry *new = &rt->dst;
2731
2732 atomic_set(&new->__refcnt, 1);
2733 new->__use = 1;
2734 new->input = dst_discard;
2735 new->output = dst_discard;
2627 dev_out = net->loopback_dev;
2628 fl.oif = dev_out->ifindex;
2629 res.fi = NULL;
2630 flags |= RTCF_LOCAL;
2631 goto make_route;
2632 }
2633
2634#ifdef CONFIG_IP_ROUTE_MULTIPATH

--- 85 unchanged lines hidden (view full) ---

2720
2721 if (rt) {
2722 struct dst_entry *new = &rt->dst;
2723
2724 atomic_set(&new->__refcnt, 1);
2725 new->__use = 1;
2726 new->input = dst_discard;
2727 new->output = dst_discard;
2736 dst_copy_metrics(new, &ort->dst);
2728 memcpy(new->metrics, ort->dst.metrics, RTAX_MAX*sizeof(u32));
2737
2738 new->dev = ort->dst.dev;
2739 if (new->dev)
2740 dev_hold(new->dev);
2741
2742 rt->fl = ort->fl;
2743
2744 rt->rt_genid = rt_genid(net);

--- 78 unchanged lines hidden (view full) ---

2823 NLA_PUT_BE32(skb, RTA_DST, rt->rt_dst);
2824
2825 if (rt->fl.fl4_src) {
2826 r->rtm_src_len = 32;
2827 NLA_PUT_BE32(skb, RTA_SRC, rt->fl.fl4_src);
2828 }
2829 if (rt->dst.dev)
2830 NLA_PUT_U32(skb, RTA_OIF, rt->dst.dev->ifindex);
2729
2730 new->dev = ort->dst.dev;
2731 if (new->dev)
2732 dev_hold(new->dev);
2733
2734 rt->fl = ort->fl;
2735
2736 rt->rt_genid = rt_genid(net);

--- 78 unchanged lines hidden (view full) ---

2815 NLA_PUT_BE32(skb, RTA_DST, rt->rt_dst);
2816
2817 if (rt->fl.fl4_src) {
2818 r->rtm_src_len = 32;
2819 NLA_PUT_BE32(skb, RTA_SRC, rt->fl.fl4_src);
2820 }
2821 if (rt->dst.dev)
2822 NLA_PUT_U32(skb, RTA_OIF, rt->dst.dev->ifindex);
2831#ifdef CONFIG_NET_CLS_ROUTE
2823#ifdef CONFIG_IP_ROUTE_CLASSID
2832 if (rt->dst.tclassid)
2833 NLA_PUT_U32(skb, RTA_FLOW, rt->dst.tclassid);
2834#endif
2835 if (rt_is_input_route(rt))
2836 NLA_PUT_BE32(skb, RTA_PREFSRC, rt->rt_spec_dst);
2837 else if (rt->rt_src != rt->fl.fl4_src)
2838 NLA_PUT_BE32(skb, RTA_PREFSRC, rt->rt_src);
2839
2840 if (rt->rt_dst != rt->rt_gateway)
2841 NLA_PUT_BE32(skb, RTA_GATEWAY, rt->rt_gateway);
2842
2824 if (rt->dst.tclassid)
2825 NLA_PUT_U32(skb, RTA_FLOW, rt->dst.tclassid);
2826#endif
2827 if (rt_is_input_route(rt))
2828 NLA_PUT_BE32(skb, RTA_PREFSRC, rt->rt_spec_dst);
2829 else if (rt->rt_src != rt->fl.fl4_src)
2830 NLA_PUT_BE32(skb, RTA_PREFSRC, rt->rt_src);
2831
2832 if (rt->rt_dst != rt->rt_gateway)
2833 NLA_PUT_BE32(skb, RTA_GATEWAY, rt->rt_gateway);
2834
2843 if (rtnetlink_put_metrics(skb, dst_metrics_ptr(&rt->dst)) < 0)
2835 if (rtnetlink_put_metrics(skb, rt->dst.metrics) < 0)
2844 goto nla_put_failure;
2845
2846 if (rt->fl.mark)
2847 NLA_PUT_BE32(skb, RTA_MARK, rt->fl.mark);
2848
2849 error = rt->dst.error;
2850 expires = rt->dst.expires ? rt->dst.expires - jiffies : 0;
2851 if (rt->peer) {

--- 95 unchanged lines hidden (view full) ---

2947 err = ip_route_input(skb, dst, src, rtm->rtm_tos, dev);
2948 local_bh_enable();
2949
2950 rt = skb_rtable(skb);
2951 if (err == 0 && rt->dst.error)
2952 err = -rt->dst.error;
2953 } else {
2954 struct flowi fl = {
2836 goto nla_put_failure;
2837
2838 if (rt->fl.mark)
2839 NLA_PUT_BE32(skb, RTA_MARK, rt->fl.mark);
2840
2841 error = rt->dst.error;
2842 expires = rt->dst.expires ? rt->dst.expires - jiffies : 0;
2843 if (rt->peer) {

--- 95 unchanged lines hidden (view full) ---

2939 err = ip_route_input(skb, dst, src, rtm->rtm_tos, dev);
2940 local_bh_enable();
2941
2942 rt = skb_rtable(skb);
2943 if (err == 0 && rt->dst.error)
2944 err = -rt->dst.error;
2945 } else {
2946 struct flowi fl = {
2955 .fl4_dst = dst,
2956 .fl4_src = src,
2957 .fl4_tos = rtm->rtm_tos,
2947 .nl_u = {
2948 .ip4_u = {
2949 .daddr = dst,
2950 .saddr = src,
2951 .tos = rtm->rtm_tos,
2952 },
2953 },
2958 .oif = tb[RTA_OIF] ? nla_get_u32(tb[RTA_OIF]) : 0,
2959 .mark = mark,
2960 };
2961 err = ip_route_output_key(net, &rt, &fl);
2962 }
2963
2964 if (err)
2965 goto errout_free;

--- 278 unchanged lines hidden (view full) ---

3244 return 0;
3245}
3246
3247static __net_initdata struct pernet_operations rt_genid_ops = {
3248 .init = rt_genid_init,
3249};
3250
3251
2954 .oif = tb[RTA_OIF] ? nla_get_u32(tb[RTA_OIF]) : 0,
2955 .mark = mark,
2956 };
2957 err = ip_route_output_key(net, &rt, &fl);
2958 }
2959
2960 if (err)
2961 goto errout_free;

--- 278 unchanged lines hidden (view full) ---

3240 return 0;
3241}
3242
3243static __net_initdata struct pernet_operations rt_genid_ops = {
3244 .init = rt_genid_init,
3245};
3246
3247
3252#ifdef CONFIG_NET_CLS_ROUTE
3248#ifdef CONFIG_IP_ROUTE_CLASSID
3253struct ip_rt_acct __percpu *ip_rt_acct __read_mostly;
3249struct ip_rt_acct __percpu *ip_rt_acct __read_mostly;
3254#endif /* CONFIG_NET_CLS_ROUTE */
3250#endif /* CONFIG_IP_ROUTE_CLASSID */
3255
3256static __initdata unsigned long rhash_entries;
3257static int __init set_rhash_entries(char *str)
3258{
3259 if (!str)
3260 return 0;
3261 rhash_entries = simple_strtoul(str, &str, 0);
3262 return 1;
3263}
3264__setup("rhash_entries=", set_rhash_entries);
3265
3266int __init ip_rt_init(void)
3267{
3268 int rc = 0;
3269
3251
3252static __initdata unsigned long rhash_entries;
3253static int __init set_rhash_entries(char *str)
3254{
3255 if (!str)
3256 return 0;
3257 rhash_entries = simple_strtoul(str, &str, 0);
3258 return 1;
3259}
3260__setup("rhash_entries=", set_rhash_entries);
3261
3262int __init ip_rt_init(void)
3263{
3264 int rc = 0;
3265
3270#ifdef CONFIG_NET_CLS_ROUTE
3266#ifdef CONFIG_IP_ROUTE_CLASSID
3271 ip_rt_acct = __alloc_percpu(256 * sizeof(struct ip_rt_acct), __alignof__(struct ip_rt_acct));
3272 if (!ip_rt_acct)
3273 panic("IP: failed to allocate ip_rt_acct\n");
3274#endif
3275
3276 ipv4_dst_ops.kmem_cachep =
3277 kmem_cache_create("ip_dst_cache", sizeof(struct rtable), 0,
3278 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);

--- 61 unchanged lines hidden --- 		3267 ip_rt_acct = __alloc_percpu(256 * sizeof(struct ip_rt_acct), __alignof__(struct ip_rt_acct));
3268 if (!ip_rt_acct)
3269 panic("IP: failed to allocate ip_rt_acct\n");
3270#endif
3271
3272 ipv4_dst_ops.kmem_cachep =
3273 kmem_cache_create("ip_dst_cache", sizeof(struct rtable), 0,
3274 SLAB_HWCACHE_ALIGN|SLAB_PANIC, NULL);

--- 61 unchanged lines hidden ---