route

how kernel add a IPv4 address

SUMMARY

There is a struct in_device __rcu *ip_ptr under every net_device.
which stores all the IPv4 related information for this device.

Each IPv4 address is stored a struct in_ifaddr,
all the IPv4 address of a net device are stored into a list(netdev->ip_ptr->ifa_list).

内核如何管理ifa

相关结构体

net_device 和in_device

1
2
3
1075 struct net_device {
...
1208         struct in_device __rcu  *ip_ptr;        /* IPv4 specific data   */
1
2
3
4
5
6
55 struct in_device {
56         struct net_device       *dev;
57         atomic_t                refcnt;
58         int                     dead;
59         struct in_ifaddr        *ifa_list;      /* IP ifaddr chain              */
...

ifa结构体struct in_ifaddr

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
161 struct in_ifaddr {
162         struct hlist_node       hash;
163         struct in_ifaddr        *ifa_next;
164         struct in_device        *ifa_dev;
165         struct rcu_head         rcu_head;
166         __be32                  ifa_local;
167         __be32                  ifa_address;
168         __be32                  ifa_mask;
169         __be32                  ifa_broadcast;
170         unsigned char           ifa_scope;
171         unsigned char           ifa_flags;
172         unsigned char           ifa_prefixlen;
173         char                    ifa_label[IFNAMSIZ];
174 
175         /* In seconds, relative to tstamp. Expiry is at tstamp + HZ * lft. */
176         __u32                   ifa_valid_lft;
177         __u32                   ifa_preferred_lft;
178         unsigned long           ifa_cstamp; /* created timestamp */
179         unsigned long           ifa_tstamp; /* updated timestamp */
180 };

注册处理增删IP地址命令的入口函数

1
2
3
4
5
2284 void __init devinet_init(void)
2285 { 
...
2300         rtnl_register(PF_INET, RTM_NEWADDR, inet_rtm_newaddr, NULL, NULL);
2301         rtnl_register(PF_INET, RTM_DELADDR, inet_rtm_deladdr, NULL, NULL);

内核如何处理增加ip地址

calltrace

1
2
3
4
5
6
7
8
> inet_rtm_newaddr
> > rtm_to_ifaddr
> > find_matching_ifa
> > __inet_insert_ifa
> > > insert ifa to ifa_list with right postion.
> > > inet_hash_insert(dev_net(in_dev->dev), ifa);
> > > rtmsg_ifa(RTM_NEWADDR, ifa, nlh, portid);
> > > blocking_notifier_call_chain(&inetaddr_chain, NETDEV_UP, ifa);

NOTE:

inetaddr_chain will be shown in next blog.

处理增加IP地址命令的入口函数

inet_rtm_newaddr

  • 创建：函数rtm_to_ifaddr根据netlink 消息，生成一个 ifa实例
  • 查找：使用函数find_matching_ifa查找是否有相同的ifa
  • 插入：如果没有找到， 则通过``__inet_insert_ifa`添加ifa到内核里。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
807 static int inet_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh)
808 {
809         struct net *net = sock_net(skb->sk);
810         struct in_ifaddr *ifa;
811         struct in_ifaddr *ifa_existing;
812         __u32 valid_lft = INFINITY_LIFE_TIME;
813         __u32 prefered_lft = INFINITY_LIFE_TIME;
814 
815         ASSERT_RTNL();
816 
817         ifa = rtm_to_ifaddr(net, nlh, &valid_lft, &prefered_lft);
818         if (IS_ERR(ifa))
819                 return PTR_ERR(ifa);
820 
821         ifa_existing = find_matching_ifa(ifa);
822         if (!ifa_existing) {
823                 /* It would be best to check for !NLM_F_CREATE here but
824                  * userspace alreay relies on not having to provide this.
825                  */
826                 set_ifa_lifetime(ifa, valid_lft, prefered_lft);
827                 return __inet_insert_ifa(ifa, nlh, NETLINK_CB(skb).portid);
828         } else {
829                 inet_free_ifa(ifa);
830 
831                 if (nlh->nlmsg_flags & NLM_F_EXCL ||
832                     !(nlh->nlmsg_flags & NLM_F_REPLACE))
833                         return -EEXIST;
834                 ifa = ifa_existing;
835                 set_ifa_lifetime(ifa, valid_lft, prefered_lft);
836                 cancel_delayed_work(&check_lifetime_work);
837                 schedule_delayed_work(&check_lifetime_work, 0);
838                 rtmsg_ifa(RTM_NEWADDR, ifa, nlh, NETLINK_CB(skb).portid);
839                 blocking_notifier_call_chain(&inetaddr_chain, NETDEV_UP, ifa);
840         }
841         return 0;
842 }
rtm_to_ifaddr

根据netlink消息生成 ifa

  • 查找并确认netdev 网口名字: 查找 ip命令指定的网口下，是否有这个IP地址，而不是在整个ns下查
  • 分配一个ifa实例
  • 根据 netlink 消息内容初始化 ifa

Q: 同一个IP地址(掩码也相同)可以配置的多个网口上吗？或者说多个网口可以配置相同的ip地址（ip 和掩码都相同）?
A: 可以

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
708 static struct in_ifaddr *rtm_to_ifaddr(struct net *net, struct nlmsghdr *nlh,
709                                        __u32 *pvalid_lft, __u32 *pprefered_lft)
710 {
711         struct nlattr *tb[IFA_MAX+1];
712         struct in_ifaddr *ifa;
713         struct ifaddrmsg *ifm;
714         struct net_device *dev;
715         struct in_device *in_dev;
716         int err;
717 
718         err = nlmsg_parse(nlh, sizeof(*ifm), tb, IFA_MAX, ifa_ipv4_policy);
719         if (err < 0)
720                 goto errout;
721 
722         ifm = nlmsg_data(nlh);
723         err = -EINVAL;
724         if (ifm->ifa_prefixlen > 32 || tb[IFA_LOCAL] == NULL)
725                 goto errout;
726 
727         dev = __dev_get_by_index(net, ifm->ifa_index);
728         err = -ENODEV;
729         if (dev == NULL)
730                 goto errout;
731 
732         in_dev = __in_dev_get_rtnl(dev);
733         err = -ENOBUFS;
734         if (in_dev == NULL)
735                 goto errout;
736 
737         ifa = inet_alloc_ifa();
738         if (ifa == NULL)
739                 /*
740                  * A potential indev allocation can be left alive, it stays
741                  * assigned to its device and is destroy with it.
742                  */
743                 goto errout;
744 
745         ipv4_devconf_setall(in_dev);
746         in_dev_hold(in_dev);
747 
748         if (tb[IFA_ADDRESS] == NULL)
749                 tb[IFA_ADDRESS] = tb[IFA_LOCAL];
750 
751         INIT_HLIST_NODE(&ifa->hash);
752         ifa->ifa_prefixlen = ifm->ifa_prefixlen;
753         ifa->ifa_mask = inet_make_mask(ifm->ifa_prefixlen);
754         ifa->ifa_flags = ifm->ifa_flags;
755         ifa->ifa_scope = ifm->ifa_scope;
756         ifa->ifa_dev = in_dev;
757 
758         ifa->ifa_local = nla_get_be32(tb[IFA_LOCAL]);
759         ifa->ifa_address = nla_get_be32(tb[IFA_ADDRESS]);
760 
761         if (tb[IFA_BROADCAST])
762                 ifa->ifa_broadcast = nla_get_be32(tb[IFA_BROADCAST]);
763 
764         if (tb[IFA_LABEL])
765                 nla_strlcpy(ifa->ifa_label, tb[IFA_LABEL], IFNAMSIZ);
766         else
767                 memcpy(ifa->ifa_label, dev->name, IFNAMSIZ);
768 
769         if (tb[IFA_CACHEINFO]) {
770                 struct ifa_cacheinfo *ci;
771 
772                 ci = nla_data(tb[IFA_CACHEINFO]);
773                 if (!ci->ifa_valid || ci->ifa_prefered > ci->ifa_valid) {
774                         err = -EINVAL;
775                         goto errout_free;
776                 }
777                 *pvalid_lft = ci->ifa_valid;
778                 *pprefered_lft = ci->ifa_prefered;
779         }
780 
781         return ifa;
782 
783 errout_free:
784         inet_free_ifa(ifa);
785 errout:
786         return ERR_PTR(err);
787 }

把 ifa 插入到对应的两个链表里。

  • 插入到网口下的 ip 地址列表
  • 插入同 ns 下的 hash 链表
  • 发送 netlink 更新消息
  • 发送blocking_notifier_call_chain :待学习
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
426 static int __inet_insert_ifa(struct in_ifaddr *ifa, struct nlmsghdr *nlh,
427                              u32 portid)
428 {
429         struct in_device *in_dev = ifa->ifa_dev;
430         struct in_ifaddr *ifa1, **ifap, **last_primary;
431 
432         ASSERT_RTNL();
433 
434         if (!ifa->ifa_local) {
435                 inet_free_ifa(ifa);
436                 return 0;
437         }
438 
439         ifa->ifa_flags &= ~IFA_F_SECONDARY;
440         last_primary = &in_dev->ifa_list;
441 
442         for (ifap = &in_dev->ifa_list; (ifa1 = *ifap) != NULL;
443              ifap = &ifa1->ifa_next) {
444                 if (!(ifa1->ifa_flags & IFA_F_SECONDARY) &&
445                     ifa->ifa_scope <= ifa1->ifa_scope)
446                         last_primary = &ifa1->ifa_next;
447                 if (ifa1->ifa_mask == ifa->ifa_mask &&
448                     inet_ifa_match(ifa1->ifa_address, ifa)) {
449                         if (ifa1->ifa_local == ifa->ifa_local) {
450                                 inet_free_ifa(ifa);
451                                 return -EEXIST;
452                         }
453                         if (ifa1->ifa_scope != ifa->ifa_scope) {
454                                 inet_free_ifa(ifa);
455                                 return -EINVAL;
456                         }
457                         ifa->ifa_flags |= IFA_F_SECONDARY;
458                 }
459         }
460 
461         if (!(ifa->ifa_flags & IFA_F_SECONDARY)) {
462                 net_srandom(ifa->ifa_local);
463                 ifap = last_primary;
464         }
465 
466         ifa->ifa_next = *ifap;
467         *ifap = ifa;
468 
469         inet_hash_insert(dev_net(in_dev->dev), ifa);
470 
471         cancel_delayed_work(&check_lifetime_work);
472         schedule_delayed_work(&check_lifetime_work, 0);
473 
474         /* Send message first, then call notifier.
475            Notifier will trigger FIB update, so that
476            listeners of netlink will know about new ifaddr */
477         rtmsg_ifa(RTM_NEWADDR, ifa, nlh, portid);
478         blocking_notifier_call_chain(&inetaddr_chain, NETDEV_UP, ifa);
479 
480         return 0;
481 }