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.
6 * Routing netlink socket interface: protocol independent part.
8 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public License
12 * as published by the Free Software Foundation; either version
13 * 2 of the License, or (at your option) any later version.
16 * Vitaly E. Lavrov RTA_OK arithmetics was wrong.
19 #include <linux/errno.h>
20 #include <linux/module.h>
21 #include <linux/types.h>
22 #include <linux/socket.h>
23 #include <linux/kernel.h>
24 #include <linux/timer.h>
25 #include <linux/string.h>
26 #include <linux/sockios.h>
27 #include <linux/net.h>
28 #include <linux/fcntl.h>
30 #include <linux/slab.h>
31 #include <linux/interrupt.h>
32 #include <linux/capability.h>
33 #include <linux/skbuff.h>
34 #include <linux/init.h>
35 #include <linux/security.h>
36 #include <linux/mutex.h>
37 #include <linux/if_addr.h>
38 #include <linux/if_bridge.h>
39 #include <linux/if_vlan.h>
40 #include <linux/pci.h>
41 #include <linux/etherdevice.h>
43 #include <asm/uaccess.h>
45 #include <linux/inet.h>
46 #include <linux/netdevice.h>
47 #include <net/switchdev.h>
49 #include <net/protocol.h>
51 #include <net/route.h>
55 #include <net/pkt_sched.h>
56 #include <net/fib_rules.h>
57 #include <net/rtnetlink.h>
58 #include <net/net_namespace.h>
62 rtnl_dumpit_func dumpit
;
63 rtnl_calcit_func calcit
;
66 static DEFINE_MUTEX(rtnl_mutex
);
70 mutex_lock(&rtnl_mutex
);
72 EXPORT_SYMBOL(rtnl_lock
);
74 void __rtnl_unlock(void)
76 mutex_unlock(&rtnl_mutex
);
79 void rtnl_unlock(void)
81 /* This fellow will unlock it for us. */
84 EXPORT_SYMBOL(rtnl_unlock
);
86 int rtnl_trylock(void)
88 return mutex_trylock(&rtnl_mutex
);
90 EXPORT_SYMBOL(rtnl_trylock
);
92 int rtnl_is_locked(void)
94 return mutex_is_locked(&rtnl_mutex
);
96 EXPORT_SYMBOL(rtnl_is_locked
);
98 #ifdef CONFIG_PROVE_LOCKING
99 int lockdep_rtnl_is_held(void)
101 return lockdep_is_held(&rtnl_mutex
);
103 EXPORT_SYMBOL(lockdep_rtnl_is_held
);
104 #endif /* #ifdef CONFIG_PROVE_LOCKING */
106 static struct rtnl_link
*rtnl_msg_handlers
[RTNL_FAMILY_MAX
+ 1];
108 static inline int rtm_msgindex(int msgtype
)
110 int msgindex
= msgtype
- RTM_BASE
;
113 * msgindex < 0 implies someone tried to register a netlink
114 * control code. msgindex >= RTM_NR_MSGTYPES may indicate that
115 * the message type has not been added to linux/rtnetlink.h
117 BUG_ON(msgindex
< 0 || msgindex
>= RTM_NR_MSGTYPES
);
122 static rtnl_doit_func
rtnl_get_doit(int protocol
, int msgindex
)
124 struct rtnl_link
*tab
;
126 if (protocol
<= RTNL_FAMILY_MAX
)
127 tab
= rtnl_msg_handlers
[protocol
];
131 if (tab
== NULL
|| tab
[msgindex
].doit
== NULL
)
132 tab
= rtnl_msg_handlers
[PF_UNSPEC
];
134 return tab
[msgindex
].doit
;
137 static rtnl_dumpit_func
rtnl_get_dumpit(int protocol
, int msgindex
)
139 struct rtnl_link
*tab
;
141 if (protocol
<= RTNL_FAMILY_MAX
)
142 tab
= rtnl_msg_handlers
[protocol
];
146 if (tab
== NULL
|| tab
[msgindex
].dumpit
== NULL
)
147 tab
= rtnl_msg_handlers
[PF_UNSPEC
];
149 return tab
[msgindex
].dumpit
;
152 static rtnl_calcit_func
rtnl_get_calcit(int protocol
, int msgindex
)
154 struct rtnl_link
*tab
;
156 if (protocol
<= RTNL_FAMILY_MAX
)
157 tab
= rtnl_msg_handlers
[protocol
];
161 if (tab
== NULL
|| tab
[msgindex
].calcit
== NULL
)
162 tab
= rtnl_msg_handlers
[PF_UNSPEC
];
164 return tab
[msgindex
].calcit
;
168 * __rtnl_register - Register a rtnetlink message type
169 * @protocol: Protocol family or PF_UNSPEC
170 * @msgtype: rtnetlink message type
171 * @doit: Function pointer called for each request message
172 * @dumpit: Function pointer called for each dump request (NLM_F_DUMP) message
173 * @calcit: Function pointer to calc size of dump message
175 * Registers the specified function pointers (at least one of them has
176 * to be non-NULL) to be called whenever a request message for the
177 * specified protocol family and message type is received.
179 * The special protocol family PF_UNSPEC may be used to define fallback
180 * function pointers for the case when no entry for the specific protocol
183 * Returns 0 on success or a negative error code.
185 int __rtnl_register(int protocol
, int msgtype
,
186 rtnl_doit_func doit
, rtnl_dumpit_func dumpit
,
187 rtnl_calcit_func calcit
)
189 struct rtnl_link
*tab
;
192 BUG_ON(protocol
< 0 || protocol
> RTNL_FAMILY_MAX
);
193 msgindex
= rtm_msgindex(msgtype
);
195 tab
= rtnl_msg_handlers
[protocol
];
197 tab
= kcalloc(RTM_NR_MSGTYPES
, sizeof(*tab
), GFP_KERNEL
);
201 rtnl_msg_handlers
[protocol
] = tab
;
205 tab
[msgindex
].doit
= doit
;
208 tab
[msgindex
].dumpit
= dumpit
;
211 tab
[msgindex
].calcit
= calcit
;
215 EXPORT_SYMBOL_GPL(__rtnl_register
);
218 * rtnl_register - Register a rtnetlink message type
220 * Identical to __rtnl_register() but panics on failure. This is useful
221 * as failure of this function is very unlikely, it can only happen due
222 * to lack of memory when allocating the chain to store all message
223 * handlers for a protocol. Meant for use in init functions where lack
224 * of memory implies no sense in continuing.
226 void rtnl_register(int protocol
, int msgtype
,
227 rtnl_doit_func doit
, rtnl_dumpit_func dumpit
,
228 rtnl_calcit_func calcit
)
230 if (__rtnl_register(protocol
, msgtype
, doit
, dumpit
, calcit
) < 0)
231 panic("Unable to register rtnetlink message handler, "
232 "protocol = %d, message type = %d\n",
235 EXPORT_SYMBOL_GPL(rtnl_register
);
238 * rtnl_unregister - Unregister a rtnetlink message type
239 * @protocol: Protocol family or PF_UNSPEC
240 * @msgtype: rtnetlink message type
242 * Returns 0 on success or a negative error code.
244 int rtnl_unregister(int protocol
, int msgtype
)
248 BUG_ON(protocol
< 0 || protocol
> RTNL_FAMILY_MAX
);
249 msgindex
= rtm_msgindex(msgtype
);
251 if (rtnl_msg_handlers
[protocol
] == NULL
)
254 rtnl_msg_handlers
[protocol
][msgindex
].doit
= NULL
;
255 rtnl_msg_handlers
[protocol
][msgindex
].dumpit
= NULL
;
259 EXPORT_SYMBOL_GPL(rtnl_unregister
);
262 * rtnl_unregister_all - Unregister all rtnetlink message type of a protocol
263 * @protocol : Protocol family or PF_UNSPEC
265 * Identical to calling rtnl_unregster() for all registered message types
266 * of a certain protocol family.
268 void rtnl_unregister_all(int protocol
)
270 BUG_ON(protocol
< 0 || protocol
> RTNL_FAMILY_MAX
);
272 kfree(rtnl_msg_handlers
[protocol
]);
273 rtnl_msg_handlers
[protocol
] = NULL
;
275 EXPORT_SYMBOL_GPL(rtnl_unregister_all
);
277 static LIST_HEAD(link_ops
);
279 static const struct rtnl_link_ops
*rtnl_link_ops_get(const char *kind
)
281 const struct rtnl_link_ops
*ops
;
283 list_for_each_entry(ops
, &link_ops
, list
) {
284 if (!strcmp(ops
->kind
, kind
))
291 * __rtnl_link_register - Register rtnl_link_ops with rtnetlink.
292 * @ops: struct rtnl_link_ops * to register
294 * The caller must hold the rtnl_mutex. This function should be used
295 * by drivers that create devices during module initialization. It
296 * must be called before registering the devices.
298 * Returns 0 on success or a negative error code.
300 int __rtnl_link_register(struct rtnl_link_ops
*ops
)
302 if (rtnl_link_ops_get(ops
->kind
))
305 /* The check for setup is here because if ops
306 * does not have that filled up, it is not possible
307 * to use the ops for creating device. So do not
308 * fill up dellink as well. That disables rtnl_dellink.
310 if (ops
->setup
&& !ops
->dellink
)
311 ops
->dellink
= unregister_netdevice_queue
;
313 list_add_tail(&ops
->list
, &link_ops
);
316 EXPORT_SYMBOL_GPL(__rtnl_link_register
);
319 * rtnl_link_register - Register rtnl_link_ops with rtnetlink.
320 * @ops: struct rtnl_link_ops * to register
322 * Returns 0 on success or a negative error code.
324 int rtnl_link_register(struct rtnl_link_ops
*ops
)
329 err
= __rtnl_link_register(ops
);
333 EXPORT_SYMBOL_GPL(rtnl_link_register
);
335 static void __rtnl_kill_links(struct net
*net
, struct rtnl_link_ops
*ops
)
337 struct net_device
*dev
;
338 LIST_HEAD(list_kill
);
340 for_each_netdev(net
, dev
) {
341 if (dev
->rtnl_link_ops
== ops
)
342 ops
->dellink(dev
, &list_kill
);
344 unregister_netdevice_many(&list_kill
);
348 * __rtnl_link_unregister - Unregister rtnl_link_ops from rtnetlink.
349 * @ops: struct rtnl_link_ops * to unregister
351 * The caller must hold the rtnl_mutex.
353 void __rtnl_link_unregister(struct rtnl_link_ops
*ops
)
358 __rtnl_kill_links(net
, ops
);
360 list_del(&ops
->list
);
362 EXPORT_SYMBOL_GPL(__rtnl_link_unregister
);
364 /* Return with the rtnl_lock held when there are no network
365 * devices unregistering in any network namespace.
367 static void rtnl_lock_unregistering_all(void)
371 DEFINE_WAIT_FUNC(wait
, woken_wake_function
);
373 add_wait_queue(&netdev_unregistering_wq
, &wait
);
375 unregistering
= false;
378 if (net
->dev_unreg_count
> 0) {
379 unregistering
= true;
387 wait_woken(&wait
, TASK_UNINTERRUPTIBLE
, MAX_SCHEDULE_TIMEOUT
);
389 remove_wait_queue(&netdev_unregistering_wq
, &wait
);
393 * rtnl_link_unregister - Unregister rtnl_link_ops from rtnetlink.
394 * @ops: struct rtnl_link_ops * to unregister
396 void rtnl_link_unregister(struct rtnl_link_ops
*ops
)
398 /* Close the race with cleanup_net() */
399 mutex_lock(&net_mutex
);
400 rtnl_lock_unregistering_all();
401 __rtnl_link_unregister(ops
);
403 mutex_unlock(&net_mutex
);
405 EXPORT_SYMBOL_GPL(rtnl_link_unregister
);
407 static size_t rtnl_link_get_slave_info_data_size(const struct net_device
*dev
)
409 struct net_device
*master_dev
;
410 const struct rtnl_link_ops
*ops
;
412 master_dev
= netdev_master_upper_dev_get((struct net_device
*) dev
);
415 ops
= master_dev
->rtnl_link_ops
;
416 if (!ops
|| !ops
->get_slave_size
)
418 /* IFLA_INFO_SLAVE_DATA + nested data */
419 return nla_total_size(sizeof(struct nlattr
)) +
420 ops
->get_slave_size(master_dev
, dev
);
423 static size_t rtnl_link_get_size(const struct net_device
*dev
)
425 const struct rtnl_link_ops
*ops
= dev
->rtnl_link_ops
;
431 size
= nla_total_size(sizeof(struct nlattr
)) + /* IFLA_LINKINFO */
432 nla_total_size(strlen(ops
->kind
) + 1); /* IFLA_INFO_KIND */
435 /* IFLA_INFO_DATA + nested data */
436 size
+= nla_total_size(sizeof(struct nlattr
)) +
439 if (ops
->get_xstats_size
)
440 /* IFLA_INFO_XSTATS */
441 size
+= nla_total_size(ops
->get_xstats_size(dev
));
443 size
+= rtnl_link_get_slave_info_data_size(dev
);
448 static LIST_HEAD(rtnl_af_ops
);
450 static const struct rtnl_af_ops
*rtnl_af_lookup(const int family
)
452 const struct rtnl_af_ops
*ops
;
454 list_for_each_entry(ops
, &rtnl_af_ops
, list
) {
455 if (ops
->family
== family
)
463 * rtnl_af_register - Register rtnl_af_ops with rtnetlink.
464 * @ops: struct rtnl_af_ops * to register
466 * Returns 0 on success or a negative error code.
468 void rtnl_af_register(struct rtnl_af_ops
*ops
)
471 list_add_tail(&ops
->list
, &rtnl_af_ops
);
474 EXPORT_SYMBOL_GPL(rtnl_af_register
);
477 * __rtnl_af_unregister - Unregister rtnl_af_ops from rtnetlink.
478 * @ops: struct rtnl_af_ops * to unregister
480 * The caller must hold the rtnl_mutex.
482 void __rtnl_af_unregister(struct rtnl_af_ops
*ops
)
484 list_del(&ops
->list
);
486 EXPORT_SYMBOL_GPL(__rtnl_af_unregister
);
489 * rtnl_af_unregister - Unregister rtnl_af_ops from rtnetlink.
490 * @ops: struct rtnl_af_ops * to unregister
492 void rtnl_af_unregister(struct rtnl_af_ops
*ops
)
495 __rtnl_af_unregister(ops
);
498 EXPORT_SYMBOL_GPL(rtnl_af_unregister
);
500 static size_t rtnl_link_get_af_size(const struct net_device
*dev
)
502 struct rtnl_af_ops
*af_ops
;
506 size
= nla_total_size(sizeof(struct nlattr
));
508 list_for_each_entry(af_ops
, &rtnl_af_ops
, list
) {
509 if (af_ops
->get_link_af_size
) {
510 /* AF_* + nested data */
511 size
+= nla_total_size(sizeof(struct nlattr
)) +
512 af_ops
->get_link_af_size(dev
);
519 static bool rtnl_have_link_slave_info(const struct net_device
*dev
)
521 struct net_device
*master_dev
;
523 master_dev
= netdev_master_upper_dev_get((struct net_device
*) dev
);
524 if (master_dev
&& master_dev
->rtnl_link_ops
)
529 static int rtnl_link_slave_info_fill(struct sk_buff
*skb
,
530 const struct net_device
*dev
)
532 struct net_device
*master_dev
;
533 const struct rtnl_link_ops
*ops
;
534 struct nlattr
*slave_data
;
537 master_dev
= netdev_master_upper_dev_get((struct net_device
*) dev
);
540 ops
= master_dev
->rtnl_link_ops
;
543 if (nla_put_string(skb
, IFLA_INFO_SLAVE_KIND
, ops
->kind
) < 0)
545 if (ops
->fill_slave_info
) {
546 slave_data
= nla_nest_start(skb
, IFLA_INFO_SLAVE_DATA
);
549 err
= ops
->fill_slave_info(skb
, master_dev
, dev
);
551 goto err_cancel_slave_data
;
552 nla_nest_end(skb
, slave_data
);
556 err_cancel_slave_data
:
557 nla_nest_cancel(skb
, slave_data
);
561 static int rtnl_link_info_fill(struct sk_buff
*skb
,
562 const struct net_device
*dev
)
564 const struct rtnl_link_ops
*ops
= dev
->rtnl_link_ops
;
570 if (nla_put_string(skb
, IFLA_INFO_KIND
, ops
->kind
) < 0)
572 if (ops
->fill_xstats
) {
573 err
= ops
->fill_xstats(skb
, dev
);
577 if (ops
->fill_info
) {
578 data
= nla_nest_start(skb
, IFLA_INFO_DATA
);
581 err
= ops
->fill_info(skb
, dev
);
583 goto err_cancel_data
;
584 nla_nest_end(skb
, data
);
589 nla_nest_cancel(skb
, data
);
593 static int rtnl_link_fill(struct sk_buff
*skb
, const struct net_device
*dev
)
595 struct nlattr
*linkinfo
;
598 linkinfo
= nla_nest_start(skb
, IFLA_LINKINFO
);
599 if (linkinfo
== NULL
)
602 err
= rtnl_link_info_fill(skb
, dev
);
604 goto err_cancel_link
;
606 err
= rtnl_link_slave_info_fill(skb
, dev
);
608 goto err_cancel_link
;
610 nla_nest_end(skb
, linkinfo
);
614 nla_nest_cancel(skb
, linkinfo
);
619 int rtnetlink_send(struct sk_buff
*skb
, struct net
*net
, u32 pid
, unsigned int group
, int echo
)
621 struct sock
*rtnl
= net
->rtnl
;
624 NETLINK_CB(skb
).dst_group
= group
;
626 atomic_inc(&skb
->users
);
627 netlink_broadcast(rtnl
, skb
, pid
, group
, GFP_KERNEL
);
629 err
= netlink_unicast(rtnl
, skb
, pid
, MSG_DONTWAIT
);
633 int rtnl_unicast(struct sk_buff
*skb
, struct net
*net
, u32 pid
)
635 struct sock
*rtnl
= net
->rtnl
;
637 return nlmsg_unicast(rtnl
, skb
, pid
);
639 EXPORT_SYMBOL(rtnl_unicast
);
641 void rtnl_notify(struct sk_buff
*skb
, struct net
*net
, u32 pid
, u32 group
,
642 struct nlmsghdr
*nlh
, gfp_t flags
)
644 struct sock
*rtnl
= net
->rtnl
;
648 report
= nlmsg_report(nlh
);
650 nlmsg_notify(rtnl
, skb
, pid
, group
, report
, flags
);
652 EXPORT_SYMBOL(rtnl_notify
);
654 void rtnl_set_sk_err(struct net
*net
, u32 group
, int error
)
656 struct sock
*rtnl
= net
->rtnl
;
658 netlink_set_err(rtnl
, 0, group
, error
);
660 EXPORT_SYMBOL(rtnl_set_sk_err
);
662 int rtnetlink_put_metrics(struct sk_buff
*skb
, u32
*metrics
)
667 mx
= nla_nest_start(skb
, RTA_METRICS
);
671 for (i
= 0; i
< RTAX_MAX
; i
++) {
673 if (i
== RTAX_CC_ALGO
- 1) {
674 char tmp
[TCP_CA_NAME_MAX
], *name
;
676 name
= tcp_ca_get_name_by_key(metrics
[i
], tmp
);
679 if (nla_put_string(skb
, i
+ 1, name
))
680 goto nla_put_failure
;
682 if (nla_put_u32(skb
, i
+ 1, metrics
[i
]))
683 goto nla_put_failure
;
690 nla_nest_cancel(skb
, mx
);
694 return nla_nest_end(skb
, mx
);
697 nla_nest_cancel(skb
, mx
);
700 EXPORT_SYMBOL(rtnetlink_put_metrics
);
702 int rtnl_put_cacheinfo(struct sk_buff
*skb
, struct dst_entry
*dst
, u32 id
,
703 long expires
, u32 error
)
705 struct rta_cacheinfo ci
= {
706 .rta_lastuse
= jiffies_delta_to_clock_t(jiffies
- dst
->lastuse
),
707 .rta_used
= dst
->__use
,
708 .rta_clntref
= atomic_read(&(dst
->__refcnt
)),
716 clock
= jiffies_to_clock_t(abs(expires
));
717 clock
= min_t(unsigned long, clock
, INT_MAX
);
718 ci
.rta_expires
= (expires
> 0) ? clock
: -clock
;
720 return nla_put(skb
, RTA_CACHEINFO
, sizeof(ci
), &ci
);
722 EXPORT_SYMBOL_GPL(rtnl_put_cacheinfo
);
724 static void set_operstate(struct net_device
*dev
, unsigned char transition
)
726 unsigned char operstate
= dev
->operstate
;
728 switch (transition
) {
730 if ((operstate
== IF_OPER_DORMANT
||
731 operstate
== IF_OPER_UNKNOWN
) &&
733 operstate
= IF_OPER_UP
;
736 case IF_OPER_DORMANT
:
737 if (operstate
== IF_OPER_UP
||
738 operstate
== IF_OPER_UNKNOWN
)
739 operstate
= IF_OPER_DORMANT
;
743 if (dev
->operstate
!= operstate
) {
744 write_lock_bh(&dev_base_lock
);
745 dev
->operstate
= operstate
;
746 write_unlock_bh(&dev_base_lock
);
747 netdev_state_change(dev
);
751 static unsigned int rtnl_dev_get_flags(const struct net_device
*dev
)
753 return (dev
->flags
& ~(IFF_PROMISC
| IFF_ALLMULTI
)) |
754 (dev
->gflags
& (IFF_PROMISC
| IFF_ALLMULTI
));
757 static unsigned int rtnl_dev_combine_flags(const struct net_device
*dev
,
758 const struct ifinfomsg
*ifm
)
760 unsigned int flags
= ifm
->ifi_flags
;
762 /* bugwards compatibility: ifi_change == 0 is treated as ~0 */
764 flags
= (flags
& ifm
->ifi_change
) |
765 (rtnl_dev_get_flags(dev
) & ~ifm
->ifi_change
);
770 static void copy_rtnl_link_stats(struct rtnl_link_stats
*a
,
771 const struct rtnl_link_stats64
*b
)
773 a
->rx_packets
= b
->rx_packets
;
774 a
->tx_packets
= b
->tx_packets
;
775 a
->rx_bytes
= b
->rx_bytes
;
776 a
->tx_bytes
= b
->tx_bytes
;
777 a
->rx_errors
= b
->rx_errors
;
778 a
->tx_errors
= b
->tx_errors
;
779 a
->rx_dropped
= b
->rx_dropped
;
780 a
->tx_dropped
= b
->tx_dropped
;
782 a
->multicast
= b
->multicast
;
783 a
->collisions
= b
->collisions
;
785 a
->rx_length_errors
= b
->rx_length_errors
;
786 a
->rx_over_errors
= b
->rx_over_errors
;
787 a
->rx_crc_errors
= b
->rx_crc_errors
;
788 a
->rx_frame_errors
= b
->rx_frame_errors
;
789 a
->rx_fifo_errors
= b
->rx_fifo_errors
;
790 a
->rx_missed_errors
= b
->rx_missed_errors
;
792 a
->tx_aborted_errors
= b
->tx_aborted_errors
;
793 a
->tx_carrier_errors
= b
->tx_carrier_errors
;
794 a
->tx_fifo_errors
= b
->tx_fifo_errors
;
795 a
->tx_heartbeat_errors
= b
->tx_heartbeat_errors
;
796 a
->tx_window_errors
= b
->tx_window_errors
;
798 a
->rx_compressed
= b
->rx_compressed
;
799 a
->tx_compressed
= b
->tx_compressed
;
802 static void copy_rtnl_link_stats64(void *v
, const struct rtnl_link_stats64
*b
)
804 memcpy(v
, b
, sizeof(*b
));
808 static inline int rtnl_vfinfo_size(const struct net_device
*dev
,
811 if (dev
->dev
.parent
&& dev_is_pci(dev
->dev
.parent
) &&
812 (ext_filter_mask
& RTEXT_FILTER_VF
)) {
813 int num_vfs
= dev_num_vf(dev
->dev
.parent
);
814 size_t size
= nla_total_size(sizeof(struct nlattr
));
815 size
+= nla_total_size(num_vfs
* sizeof(struct nlattr
));
817 (nla_total_size(sizeof(struct ifla_vf_mac
)) +
818 nla_total_size(sizeof(struct ifla_vf_vlan
)) +
819 nla_total_size(sizeof(struct ifla_vf_spoofchk
)) +
820 nla_total_size(sizeof(struct ifla_vf_rate
)) +
821 nla_total_size(sizeof(struct ifla_vf_link_state
)) +
822 nla_total_size(sizeof(struct ifla_vf_rss_query_en
)));
828 static size_t rtnl_port_size(const struct net_device
*dev
,
831 size_t port_size
= nla_total_size(4) /* PORT_VF */
832 + nla_total_size(PORT_PROFILE_MAX
) /* PORT_PROFILE */
833 + nla_total_size(sizeof(struct ifla_port_vsi
))
835 + nla_total_size(PORT_UUID_MAX
) /* PORT_INSTANCE_UUID */
836 + nla_total_size(PORT_UUID_MAX
) /* PORT_HOST_UUID */
837 + nla_total_size(1) /* PROT_VDP_REQUEST */
838 + nla_total_size(2); /* PORT_VDP_RESPONSE */
839 size_t vf_ports_size
= nla_total_size(sizeof(struct nlattr
));
840 size_t vf_port_size
= nla_total_size(sizeof(struct nlattr
))
842 size_t port_self_size
= nla_total_size(sizeof(struct nlattr
))
845 if (!dev
->netdev_ops
->ndo_get_vf_port
|| !dev
->dev
.parent
||
846 !(ext_filter_mask
& RTEXT_FILTER_VF
))
848 if (dev_num_vf(dev
->dev
.parent
))
849 return port_self_size
+ vf_ports_size
+
850 vf_port_size
* dev_num_vf(dev
->dev
.parent
);
852 return port_self_size
;
855 static noinline
size_t if_nlmsg_size(const struct net_device
*dev
,
858 return NLMSG_ALIGN(sizeof(struct ifinfomsg
))
859 + nla_total_size(IFNAMSIZ
) /* IFLA_IFNAME */
860 + nla_total_size(IFALIASZ
) /* IFLA_IFALIAS */
861 + nla_total_size(IFNAMSIZ
) /* IFLA_QDISC */
862 + nla_total_size(sizeof(struct rtnl_link_ifmap
))
863 + nla_total_size(sizeof(struct rtnl_link_stats
))
864 + nla_total_size(sizeof(struct rtnl_link_stats64
))
865 + nla_total_size(MAX_ADDR_LEN
) /* IFLA_ADDRESS */
866 + nla_total_size(MAX_ADDR_LEN
) /* IFLA_BROADCAST */
867 + nla_total_size(4) /* IFLA_TXQLEN */
868 + nla_total_size(4) /* IFLA_WEIGHT */
869 + nla_total_size(4) /* IFLA_MTU */
870 + nla_total_size(4) /* IFLA_LINK */
871 + nla_total_size(4) /* IFLA_MASTER */
872 + nla_total_size(1) /* IFLA_CARRIER */
873 + nla_total_size(4) /* IFLA_PROMISCUITY */
874 + nla_total_size(4) /* IFLA_NUM_TX_QUEUES */
875 + nla_total_size(4) /* IFLA_NUM_RX_QUEUES */
876 + nla_total_size(1) /* IFLA_OPERSTATE */
877 + nla_total_size(1) /* IFLA_LINKMODE */
878 + nla_total_size(4) /* IFLA_CARRIER_CHANGES */
879 + nla_total_size(4) /* IFLA_LINK_NETNSID */
880 + nla_total_size(ext_filter_mask
881 & RTEXT_FILTER_VF
? 4 : 0) /* IFLA_NUM_VF */
882 + rtnl_vfinfo_size(dev
, ext_filter_mask
) /* IFLA_VFINFO_LIST */
883 + rtnl_port_size(dev
, ext_filter_mask
) /* IFLA_VF_PORTS + IFLA_PORT_SELF */
884 + rtnl_link_get_size(dev
) /* IFLA_LINKINFO */
885 + rtnl_link_get_af_size(dev
) /* IFLA_AF_SPEC */
886 + nla_total_size(MAX_PHYS_ITEM_ID_LEN
) /* IFLA_PHYS_PORT_ID */
887 + nla_total_size(MAX_PHYS_ITEM_ID_LEN
); /* IFLA_PHYS_SWITCH_ID */
890 static int rtnl_vf_ports_fill(struct sk_buff
*skb
, struct net_device
*dev
)
892 struct nlattr
*vf_ports
;
893 struct nlattr
*vf_port
;
897 vf_ports
= nla_nest_start(skb
, IFLA_VF_PORTS
);
901 for (vf
= 0; vf
< dev_num_vf(dev
->dev
.parent
); vf
++) {
902 vf_port
= nla_nest_start(skb
, IFLA_VF_PORT
);
904 goto nla_put_failure
;
905 if (nla_put_u32(skb
, IFLA_PORT_VF
, vf
))
906 goto nla_put_failure
;
907 err
= dev
->netdev_ops
->ndo_get_vf_port(dev
, vf
, skb
);
908 if (err
== -EMSGSIZE
)
909 goto nla_put_failure
;
911 nla_nest_cancel(skb
, vf_port
);
914 nla_nest_end(skb
, vf_port
);
917 nla_nest_end(skb
, vf_ports
);
922 nla_nest_cancel(skb
, vf_ports
);
926 static int rtnl_port_self_fill(struct sk_buff
*skb
, struct net_device
*dev
)
928 struct nlattr
*port_self
;
931 port_self
= nla_nest_start(skb
, IFLA_PORT_SELF
);
935 err
= dev
->netdev_ops
->ndo_get_vf_port(dev
, PORT_SELF_VF
, skb
);
937 nla_nest_cancel(skb
, port_self
);
938 return (err
== -EMSGSIZE
) ? err
: 0;
941 nla_nest_end(skb
, port_self
);
946 static int rtnl_port_fill(struct sk_buff
*skb
, struct net_device
*dev
,
951 if (!dev
->netdev_ops
->ndo_get_vf_port
|| !dev
->dev
.parent
||
952 !(ext_filter_mask
& RTEXT_FILTER_VF
))
955 err
= rtnl_port_self_fill(skb
, dev
);
959 if (dev_num_vf(dev
->dev
.parent
)) {
960 err
= rtnl_vf_ports_fill(skb
, dev
);
968 static int rtnl_phys_port_id_fill(struct sk_buff
*skb
, struct net_device
*dev
)
971 struct netdev_phys_item_id ppid
;
973 err
= dev_get_phys_port_id(dev
, &ppid
);
975 if (err
== -EOPNOTSUPP
)
980 if (nla_put(skb
, IFLA_PHYS_PORT_ID
, ppid
.id_len
, ppid
.id
))
986 static int rtnl_phys_port_name_fill(struct sk_buff
*skb
, struct net_device
*dev
)
991 err
= dev_get_phys_port_name(dev
, name
, sizeof(name
));
993 if (err
== -EOPNOTSUPP
)
998 if (nla_put(skb
, IFLA_PHYS_PORT_NAME
, strlen(name
), name
))
1004 static int rtnl_phys_switch_id_fill(struct sk_buff
*skb
, struct net_device
*dev
)
1007 struct netdev_phys_item_id psid
;
1009 err
= netdev_switch_parent_id_get(dev
, &psid
);
1011 if (err
== -EOPNOTSUPP
)
1016 if (nla_put(skb
, IFLA_PHYS_SWITCH_ID
, psid
.id_len
, psid
.id
))
1022 static int rtnl_fill_ifinfo(struct sk_buff
*skb
, struct net_device
*dev
,
1023 int type
, u32 pid
, u32 seq
, u32 change
,
1024 unsigned int flags
, u32 ext_filter_mask
)
1026 struct ifinfomsg
*ifm
;
1027 struct nlmsghdr
*nlh
;
1028 struct rtnl_link_stats64 temp
;
1029 const struct rtnl_link_stats64
*stats
;
1030 struct nlattr
*attr
, *af_spec
;
1031 struct rtnl_af_ops
*af_ops
;
1032 struct net_device
*upper_dev
= netdev_master_upper_dev_get(dev
);
1035 nlh
= nlmsg_put(skb
, pid
, seq
, type
, sizeof(*ifm
), flags
);
1039 ifm
= nlmsg_data(nlh
);
1040 ifm
->ifi_family
= AF_UNSPEC
;
1042 ifm
->ifi_type
= dev
->type
;
1043 ifm
->ifi_index
= dev
->ifindex
;
1044 ifm
->ifi_flags
= dev_get_flags(dev
);
1045 ifm
->ifi_change
= change
;
1047 if (nla_put_string(skb
, IFLA_IFNAME
, dev
->name
) ||
1048 nla_put_u32(skb
, IFLA_TXQLEN
, dev
->tx_queue_len
) ||
1049 nla_put_u8(skb
, IFLA_OPERSTATE
,
1050 netif_running(dev
) ? dev
->operstate
: IF_OPER_DOWN
) ||
1051 nla_put_u8(skb
, IFLA_LINKMODE
, dev
->link_mode
) ||
1052 nla_put_u32(skb
, IFLA_MTU
, dev
->mtu
) ||
1053 nla_put_u32(skb
, IFLA_GROUP
, dev
->group
) ||
1054 nla_put_u32(skb
, IFLA_PROMISCUITY
, dev
->promiscuity
) ||
1055 nla_put_u32(skb
, IFLA_NUM_TX_QUEUES
, dev
->num_tx_queues
) ||
1057 nla_put_u32(skb
, IFLA_NUM_RX_QUEUES
, dev
->num_rx_queues
) ||
1059 (dev
->ifindex
!= dev_get_iflink(dev
) &&
1060 nla_put_u32(skb
, IFLA_LINK
, dev_get_iflink(dev
))) ||
1062 nla_put_u32(skb
, IFLA_MASTER
, upper_dev
->ifindex
)) ||
1063 nla_put_u8(skb
, IFLA_CARRIER
, netif_carrier_ok(dev
)) ||
1065 nla_put_string(skb
, IFLA_QDISC
, dev
->qdisc
->ops
->id
)) ||
1067 nla_put_string(skb
, IFLA_IFALIAS
, dev
->ifalias
)) ||
1068 nla_put_u32(skb
, IFLA_CARRIER_CHANGES
,
1069 atomic_read(&dev
->carrier_changes
)))
1070 goto nla_put_failure
;
1073 struct rtnl_link_ifmap map
= {
1074 .mem_start
= dev
->mem_start
,
1075 .mem_end
= dev
->mem_end
,
1076 .base_addr
= dev
->base_addr
,
1079 .port
= dev
->if_port
,
1081 if (nla_put(skb
, IFLA_MAP
, sizeof(map
), &map
))
1082 goto nla_put_failure
;
1085 if (dev
->addr_len
) {
1086 if (nla_put(skb
, IFLA_ADDRESS
, dev
->addr_len
, dev
->dev_addr
) ||
1087 nla_put(skb
, IFLA_BROADCAST
, dev
->addr_len
, dev
->broadcast
))
1088 goto nla_put_failure
;
1091 if (rtnl_phys_port_id_fill(skb
, dev
))
1092 goto nla_put_failure
;
1094 if (rtnl_phys_port_name_fill(skb
, dev
))
1095 goto nla_put_failure
;
1097 if (rtnl_phys_switch_id_fill(skb
, dev
))
1098 goto nla_put_failure
;
1100 attr
= nla_reserve(skb
, IFLA_STATS
,
1101 sizeof(struct rtnl_link_stats
));
1103 goto nla_put_failure
;
1105 stats
= dev_get_stats(dev
, &temp
);
1106 copy_rtnl_link_stats(nla_data(attr
), stats
);
1108 attr
= nla_reserve(skb
, IFLA_STATS64
,
1109 sizeof(struct rtnl_link_stats64
));
1111 goto nla_put_failure
;
1112 copy_rtnl_link_stats64(nla_data(attr
), stats
);
1114 if (dev
->dev
.parent
&& (ext_filter_mask
& RTEXT_FILTER_VF
) &&
1115 nla_put_u32(skb
, IFLA_NUM_VF
, dev_num_vf(dev
->dev
.parent
)))
1116 goto nla_put_failure
;
1118 if (dev
->netdev_ops
->ndo_get_vf_config
&& dev
->dev
.parent
1119 && (ext_filter_mask
& RTEXT_FILTER_VF
)) {
1122 struct nlattr
*vfinfo
, *vf
;
1123 int num_vfs
= dev_num_vf(dev
->dev
.parent
);
1125 vfinfo
= nla_nest_start(skb
, IFLA_VFINFO_LIST
);
1127 goto nla_put_failure
;
1128 for (i
= 0; i
< num_vfs
; i
++) {
1129 struct ifla_vf_info ivi
;
1130 struct ifla_vf_mac vf_mac
;
1131 struct ifla_vf_vlan vf_vlan
;
1132 struct ifla_vf_rate vf_rate
;
1133 struct ifla_vf_tx_rate vf_tx_rate
;
1134 struct ifla_vf_spoofchk vf_spoofchk
;
1135 struct ifla_vf_link_state vf_linkstate
;
1136 struct ifla_vf_rss_query_en vf_rss_query_en
;
1139 * Not all SR-IOV capable drivers support the
1140 * spoofcheck and "RSS query enable" query. Preset to
1141 * -1 so the user space tool can detect that the driver
1142 * didn't report anything.
1145 ivi
.rss_query_en
= -1;
1146 memset(ivi
.mac
, 0, sizeof(ivi
.mac
));
1147 /* The default value for VF link state is "auto"
1148 * IFLA_VF_LINK_STATE_AUTO which equals zero
1151 if (dev
->netdev_ops
->ndo_get_vf_config(dev
, i
, &ivi
))
1159 vf_rss_query_en
.vf
= ivi
.vf
;
1161 memcpy(vf_mac
.mac
, ivi
.mac
, sizeof(ivi
.mac
));
1162 vf_vlan
.vlan
= ivi
.vlan
;
1163 vf_vlan
.qos
= ivi
.qos
;
1164 vf_tx_rate
.rate
= ivi
.max_tx_rate
;
1165 vf_rate
.min_tx_rate
= ivi
.min_tx_rate
;
1166 vf_rate
.max_tx_rate
= ivi
.max_tx_rate
;
1167 vf_spoofchk
.setting
= ivi
.spoofchk
;
1168 vf_linkstate
.link_state
= ivi
.linkstate
;
1169 vf_rss_query_en
.setting
= ivi
.rss_query_en
;
1170 vf
= nla_nest_start(skb
, IFLA_VF_INFO
);
1172 nla_nest_cancel(skb
, vfinfo
);
1173 goto nla_put_failure
;
1175 if (nla_put(skb
, IFLA_VF_MAC
, sizeof(vf_mac
), &vf_mac
) ||
1176 nla_put(skb
, IFLA_VF_VLAN
, sizeof(vf_vlan
), &vf_vlan
) ||
1177 nla_put(skb
, IFLA_VF_RATE
, sizeof(vf_rate
),
1179 nla_put(skb
, IFLA_VF_TX_RATE
, sizeof(vf_tx_rate
),
1181 nla_put(skb
, IFLA_VF_SPOOFCHK
, sizeof(vf_spoofchk
),
1183 nla_put(skb
, IFLA_VF_LINK_STATE
, sizeof(vf_linkstate
),
1185 nla_put(skb
, IFLA_VF_RSS_QUERY_EN
,
1186 sizeof(vf_rss_query_en
),
1188 goto nla_put_failure
;
1189 nla_nest_end(skb
, vf
);
1191 nla_nest_end(skb
, vfinfo
);
1194 if (rtnl_port_fill(skb
, dev
, ext_filter_mask
))
1195 goto nla_put_failure
;
1197 if (dev
->rtnl_link_ops
|| rtnl_have_link_slave_info(dev
)) {
1198 if (rtnl_link_fill(skb
, dev
) < 0)
1199 goto nla_put_failure
;
1202 if (dev
->rtnl_link_ops
&&
1203 dev
->rtnl_link_ops
->get_link_net
) {
1204 struct net
*link_net
= dev
->rtnl_link_ops
->get_link_net(dev
);
1206 if (!net_eq(dev_net(dev
), link_net
)) {
1207 int id
= peernet2id(dev_net(dev
), link_net
);
1209 if (nla_put_s32(skb
, IFLA_LINK_NETNSID
, id
))
1210 goto nla_put_failure
;
1214 if (!(af_spec
= nla_nest_start(skb
, IFLA_AF_SPEC
)))
1215 goto nla_put_failure
;
1217 list_for_each_entry(af_ops
, &rtnl_af_ops
, list
) {
1218 if (af_ops
->fill_link_af
) {
1222 if (!(af
= nla_nest_start(skb
, af_ops
->family
)))
1223 goto nla_put_failure
;
1225 err
= af_ops
->fill_link_af(skb
, dev
);
1228 * Caller may return ENODATA to indicate that there
1229 * was no data to be dumped. This is not an error, it
1230 * means we should trim the attribute header and
1233 if (err
== -ENODATA
)
1234 nla_nest_cancel(skb
, af
);
1236 goto nla_put_failure
;
1238 nla_nest_end(skb
, af
);
1242 nla_nest_end(skb
, af_spec
);
1244 nlmsg_end(skb
, nlh
);
1248 nlmsg_cancel(skb
, nlh
);
1252 static const struct nla_policy ifla_policy
[IFLA_MAX
+1] = {
1253 [IFLA_IFNAME
] = { .type
= NLA_STRING
, .len
= IFNAMSIZ
-1 },
1254 [IFLA_ADDRESS
] = { .type
= NLA_BINARY
, .len
= MAX_ADDR_LEN
},
1255 [IFLA_BROADCAST
] = { .type
= NLA_BINARY
, .len
= MAX_ADDR_LEN
},
1256 [IFLA_MAP
] = { .len
= sizeof(struct rtnl_link_ifmap
) },
1257 [IFLA_MTU
] = { .type
= NLA_U32
},
1258 [IFLA_LINK
] = { .type
= NLA_U32
},
1259 [IFLA_MASTER
] = { .type
= NLA_U32
},
1260 [IFLA_CARRIER
] = { .type
= NLA_U8
},
1261 [IFLA_TXQLEN
] = { .type
= NLA_U32
},
1262 [IFLA_WEIGHT
] = { .type
= NLA_U32
},
1263 [IFLA_OPERSTATE
] = { .type
= NLA_U8
},
1264 [IFLA_LINKMODE
] = { .type
= NLA_U8
},
1265 [IFLA_LINKINFO
] = { .type
= NLA_NESTED
},
1266 [IFLA_NET_NS_PID
] = { .type
= NLA_U32
},
1267 [IFLA_NET_NS_FD
] = { .type
= NLA_U32
},
1268 [IFLA_IFALIAS
] = { .type
= NLA_STRING
, .len
= IFALIASZ
-1 },
1269 [IFLA_VFINFO_LIST
] = {. type
= NLA_NESTED
},
1270 [IFLA_VF_PORTS
] = { .type
= NLA_NESTED
},
1271 [IFLA_PORT_SELF
] = { .type
= NLA_NESTED
},
1272 [IFLA_AF_SPEC
] = { .type
= NLA_NESTED
},
1273 [IFLA_EXT_MASK
] = { .type
= NLA_U32
},
1274 [IFLA_PROMISCUITY
] = { .type
= NLA_U32
},
1275 [IFLA_NUM_TX_QUEUES
] = { .type
= NLA_U32
},
1276 [IFLA_NUM_RX_QUEUES
] = { .type
= NLA_U32
},
1277 [IFLA_PHYS_PORT_ID
] = { .type
= NLA_BINARY
, .len
= MAX_PHYS_ITEM_ID_LEN
},
1278 [IFLA_CARRIER_CHANGES
] = { .type
= NLA_U32
}, /* ignored */
1279 [IFLA_PHYS_SWITCH_ID
] = { .type
= NLA_BINARY
, .len
= MAX_PHYS_ITEM_ID_LEN
},
1280 [IFLA_LINK_NETNSID
] = { .type
= NLA_S32
},
1283 static const struct nla_policy ifla_info_policy
[IFLA_INFO_MAX
+1] = {
1284 [IFLA_INFO_KIND
] = { .type
= NLA_STRING
},
1285 [IFLA_INFO_DATA
] = { .type
= NLA_NESTED
},
1286 [IFLA_INFO_SLAVE_KIND
] = { .type
= NLA_STRING
},
1287 [IFLA_INFO_SLAVE_DATA
] = { .type
= NLA_NESTED
},
1290 static const struct nla_policy ifla_vfinfo_policy
[IFLA_VF_INFO_MAX
+1] = {
1291 [IFLA_VF_INFO
] = { .type
= NLA_NESTED
},
1294 static const struct nla_policy ifla_vf_policy
[IFLA_VF_MAX
+1] = {
1295 [IFLA_VF_MAC
] = { .len
= sizeof(struct ifla_vf_mac
) },
1296 [IFLA_VF_VLAN
] = { .len
= sizeof(struct ifla_vf_vlan
) },
1297 [IFLA_VF_TX_RATE
] = { .len
= sizeof(struct ifla_vf_tx_rate
) },
1298 [IFLA_VF_SPOOFCHK
] = { .len
= sizeof(struct ifla_vf_spoofchk
) },
1299 [IFLA_VF_RATE
] = { .len
= sizeof(struct ifla_vf_rate
) },
1300 [IFLA_VF_LINK_STATE
] = { .len
= sizeof(struct ifla_vf_link_state
) },
1301 [IFLA_VF_RSS_QUERY_EN
] = { .len
= sizeof(struct ifla_vf_rss_query_en
) },
1304 static const struct nla_policy ifla_port_policy
[IFLA_PORT_MAX
+1] = {
1305 [IFLA_PORT_VF
] = { .type
= NLA_U32
},
1306 [IFLA_PORT_PROFILE
] = { .type
= NLA_STRING
,
1307 .len
= PORT_PROFILE_MAX
},
1308 [IFLA_PORT_VSI_TYPE
] = { .type
= NLA_BINARY
,
1309 .len
= sizeof(struct ifla_port_vsi
)},
1310 [IFLA_PORT_INSTANCE_UUID
] = { .type
= NLA_BINARY
,
1311 .len
= PORT_UUID_MAX
},
1312 [IFLA_PORT_HOST_UUID
] = { .type
= NLA_STRING
,
1313 .len
= PORT_UUID_MAX
},
1314 [IFLA_PORT_REQUEST
] = { .type
= NLA_U8
, },
1315 [IFLA_PORT_RESPONSE
] = { .type
= NLA_U16
, },
1318 static int rtnl_dump_ifinfo(struct sk_buff
*skb
, struct netlink_callback
*cb
)
1320 struct net
*net
= sock_net(skb
->sk
);
1323 struct net_device
*dev
;
1324 struct hlist_head
*head
;
1325 struct nlattr
*tb
[IFLA_MAX
+1];
1326 u32 ext_filter_mask
= 0;
1331 s_idx
= cb
->args
[1];
1333 cb
->seq
= net
->dev_base_seq
;
1335 /* A hack to preserve kernel<->userspace interface.
1336 * The correct header is ifinfomsg. It is consistent with rtnl_getlink.
1337 * However, before Linux v3.9 the code here assumed rtgenmsg and that's
1338 * what iproute2 < v3.9.0 used.
1339 * We can detect the old iproute2. Even including the IFLA_EXT_MASK
1340 * attribute, its netlink message is shorter than struct ifinfomsg.
1342 hdrlen
= nlmsg_len(cb
->nlh
) < sizeof(struct ifinfomsg
) ?
1343 sizeof(struct rtgenmsg
) : sizeof(struct ifinfomsg
);
1345 if (nlmsg_parse(cb
->nlh
, hdrlen
, tb
, IFLA_MAX
, ifla_policy
) >= 0) {
1347 if (tb
[IFLA_EXT_MASK
])
1348 ext_filter_mask
= nla_get_u32(tb
[IFLA_EXT_MASK
]);
1351 for (h
= s_h
; h
< NETDEV_HASHENTRIES
; h
++, s_idx
= 0) {
1353 head
= &net
->dev_index_head
[h
];
1354 hlist_for_each_entry(dev
, head
, index_hlist
) {
1357 err
= rtnl_fill_ifinfo(skb
, dev
, RTM_NEWLINK
,
1358 NETLINK_CB(cb
->skb
).portid
,
1359 cb
->nlh
->nlmsg_seq
, 0,
1362 /* If we ran out of room on the first message,
1365 WARN_ON((err
== -EMSGSIZE
) && (skb
->len
== 0));
1370 nl_dump_check_consistent(cb
, nlmsg_hdr(skb
));
1382 int rtnl_nla_parse_ifla(struct nlattr
**tb
, const struct nlattr
*head
, int len
)
1384 return nla_parse(tb
, IFLA_MAX
, head
, len
, ifla_policy
);
1386 EXPORT_SYMBOL(rtnl_nla_parse_ifla
);
1388 struct net
*rtnl_link_get_net(struct net
*src_net
, struct nlattr
*tb
[])
1391 /* Examine the link attributes and figure out which
1392 * network namespace we are talking about.
1394 if (tb
[IFLA_NET_NS_PID
])
1395 net
= get_net_ns_by_pid(nla_get_u32(tb
[IFLA_NET_NS_PID
]));
1396 else if (tb
[IFLA_NET_NS_FD
])
1397 net
= get_net_ns_by_fd(nla_get_u32(tb
[IFLA_NET_NS_FD
]));
1399 net
= get_net(src_net
);
1402 EXPORT_SYMBOL(rtnl_link_get_net
);
1404 static int validate_linkmsg(struct net_device
*dev
, struct nlattr
*tb
[])
1407 if (tb
[IFLA_ADDRESS
] &&
1408 nla_len(tb
[IFLA_ADDRESS
]) < dev
->addr_len
)
1411 if (tb
[IFLA_BROADCAST
] &&
1412 nla_len(tb
[IFLA_BROADCAST
]) < dev
->addr_len
)
1416 if (tb
[IFLA_AF_SPEC
]) {
1420 nla_for_each_nested(af
, tb
[IFLA_AF_SPEC
], rem
) {
1421 const struct rtnl_af_ops
*af_ops
;
1423 if (!(af_ops
= rtnl_af_lookup(nla_type(af
))))
1424 return -EAFNOSUPPORT
;
1426 if (!af_ops
->set_link_af
)
1429 if (af_ops
->validate_link_af
) {
1430 err
= af_ops
->validate_link_af(dev
, af
);
1440 static int do_setvfinfo(struct net_device
*dev
, struct nlattr
*attr
)
1442 int rem
, err
= -EINVAL
;
1444 const struct net_device_ops
*ops
= dev
->netdev_ops
;
1446 nla_for_each_nested(vf
, attr
, rem
) {
1447 switch (nla_type(vf
)) {
1449 struct ifla_vf_mac
*ivm
;
1452 if (ops
->ndo_set_vf_mac
)
1453 err
= ops
->ndo_set_vf_mac(dev
, ivm
->vf
,
1457 case IFLA_VF_VLAN
: {
1458 struct ifla_vf_vlan
*ivv
;
1461 if (ops
->ndo_set_vf_vlan
)
1462 err
= ops
->ndo_set_vf_vlan(dev
, ivv
->vf
,
1467 case IFLA_VF_TX_RATE
: {
1468 struct ifla_vf_tx_rate
*ivt
;
1469 struct ifla_vf_info ivf
;
1472 if (ops
->ndo_get_vf_config
)
1473 err
= ops
->ndo_get_vf_config(dev
, ivt
->vf
,
1478 if (ops
->ndo_set_vf_rate
)
1479 err
= ops
->ndo_set_vf_rate(dev
, ivt
->vf
,
1484 case IFLA_VF_RATE
: {
1485 struct ifla_vf_rate
*ivt
;
1488 if (ops
->ndo_set_vf_rate
)
1489 err
= ops
->ndo_set_vf_rate(dev
, ivt
->vf
,
1494 case IFLA_VF_SPOOFCHK
: {
1495 struct ifla_vf_spoofchk
*ivs
;
1498 if (ops
->ndo_set_vf_spoofchk
)
1499 err
= ops
->ndo_set_vf_spoofchk(dev
, ivs
->vf
,
1503 case IFLA_VF_LINK_STATE
: {
1504 struct ifla_vf_link_state
*ivl
;
1507 if (ops
->ndo_set_vf_link_state
)
1508 err
= ops
->ndo_set_vf_link_state(dev
, ivl
->vf
,
1512 case IFLA_VF_RSS_QUERY_EN
: {
1513 struct ifla_vf_rss_query_en
*ivrssq_en
;
1515 ivrssq_en
= nla_data(vf
);
1517 if (ops
->ndo_set_vf_rss_query_en
)
1518 err
= ops
->ndo_set_vf_rss_query_en(dev
,
1520 ivrssq_en
->setting
);
1533 static int do_set_master(struct net_device
*dev
, int ifindex
)
1535 struct net_device
*upper_dev
= netdev_master_upper_dev_get(dev
);
1536 const struct net_device_ops
*ops
;
1540 if (upper_dev
->ifindex
== ifindex
)
1542 ops
= upper_dev
->netdev_ops
;
1543 if (ops
->ndo_del_slave
) {
1544 err
= ops
->ndo_del_slave(upper_dev
, dev
);
1553 upper_dev
= __dev_get_by_index(dev_net(dev
), ifindex
);
1556 ops
= upper_dev
->netdev_ops
;
1557 if (ops
->ndo_add_slave
) {
1558 err
= ops
->ndo_add_slave(upper_dev
, dev
);
1568 #define DO_SETLINK_MODIFIED 0x01
1569 /* notify flag means notify + modified. */
1570 #define DO_SETLINK_NOTIFY 0x03
1571 static int do_setlink(const struct sk_buff
*skb
,
1572 struct net_device
*dev
, struct ifinfomsg
*ifm
,
1573 struct nlattr
**tb
, char *ifname
, int status
)
1575 const struct net_device_ops
*ops
= dev
->netdev_ops
;
1578 if (tb
[IFLA_NET_NS_PID
] || tb
[IFLA_NET_NS_FD
]) {
1579 struct net
*net
= rtnl_link_get_net(dev_net(dev
), tb
);
1584 if (!netlink_ns_capable(skb
, net
->user_ns
, CAP_NET_ADMIN
)) {
1589 err
= dev_change_net_namespace(dev
, net
, ifname
);
1593 status
|= DO_SETLINK_MODIFIED
;
1597 struct rtnl_link_ifmap
*u_map
;
1600 if (!ops
->ndo_set_config
) {
1605 if (!netif_device_present(dev
)) {
1610 u_map
= nla_data(tb
[IFLA_MAP
]);
1611 k_map
.mem_start
= (unsigned long) u_map
->mem_start
;
1612 k_map
.mem_end
= (unsigned long) u_map
->mem_end
;
1613 k_map
.base_addr
= (unsigned short) u_map
->base_addr
;
1614 k_map
.irq
= (unsigned char) u_map
->irq
;
1615 k_map
.dma
= (unsigned char) u_map
->dma
;
1616 k_map
.port
= (unsigned char) u_map
->port
;
1618 err
= ops
->ndo_set_config(dev
, &k_map
);
1622 status
|= DO_SETLINK_NOTIFY
;
1625 if (tb
[IFLA_ADDRESS
]) {
1626 struct sockaddr
*sa
;
1629 len
= sizeof(sa_family_t
) + dev
->addr_len
;
1630 sa
= kmalloc(len
, GFP_KERNEL
);
1635 sa
->sa_family
= dev
->type
;
1636 memcpy(sa
->sa_data
, nla_data(tb
[IFLA_ADDRESS
]),
1638 err
= dev_set_mac_address(dev
, sa
);
1642 status
|= DO_SETLINK_MODIFIED
;
1646 err
= dev_set_mtu(dev
, nla_get_u32(tb
[IFLA_MTU
]));
1649 status
|= DO_SETLINK_MODIFIED
;
1652 if (tb
[IFLA_GROUP
]) {
1653 dev_set_group(dev
, nla_get_u32(tb
[IFLA_GROUP
]));
1654 status
|= DO_SETLINK_NOTIFY
;
1658 * Interface selected by interface index but interface
1659 * name provided implies that a name change has been
1662 if (ifm
->ifi_index
> 0 && ifname
[0]) {
1663 err
= dev_change_name(dev
, ifname
);
1666 status
|= DO_SETLINK_MODIFIED
;
1669 if (tb
[IFLA_IFALIAS
]) {
1670 err
= dev_set_alias(dev
, nla_data(tb
[IFLA_IFALIAS
]),
1671 nla_len(tb
[IFLA_IFALIAS
]));
1674 status
|= DO_SETLINK_NOTIFY
;
1677 if (tb
[IFLA_BROADCAST
]) {
1678 nla_memcpy(dev
->broadcast
, tb
[IFLA_BROADCAST
], dev
->addr_len
);
1679 call_netdevice_notifiers(NETDEV_CHANGEADDR
, dev
);
1682 if (ifm
->ifi_flags
|| ifm
->ifi_change
) {
1683 err
= dev_change_flags(dev
, rtnl_dev_combine_flags(dev
, ifm
));
1688 if (tb
[IFLA_MASTER
]) {
1689 err
= do_set_master(dev
, nla_get_u32(tb
[IFLA_MASTER
]));
1692 status
|= DO_SETLINK_MODIFIED
;
1695 if (tb
[IFLA_CARRIER
]) {
1696 err
= dev_change_carrier(dev
, nla_get_u8(tb
[IFLA_CARRIER
]));
1699 status
|= DO_SETLINK_MODIFIED
;
1702 if (tb
[IFLA_TXQLEN
]) {
1703 unsigned long value
= nla_get_u32(tb
[IFLA_TXQLEN
]);
1705 if (dev
->tx_queue_len
^ value
)
1706 status
|= DO_SETLINK_NOTIFY
;
1708 dev
->tx_queue_len
= value
;
1711 if (tb
[IFLA_OPERSTATE
])
1712 set_operstate(dev
, nla_get_u8(tb
[IFLA_OPERSTATE
]));
1714 if (tb
[IFLA_LINKMODE
]) {
1715 unsigned char value
= nla_get_u8(tb
[IFLA_LINKMODE
]);
1717 write_lock_bh(&dev_base_lock
);
1718 if (dev
->link_mode
^ value
)
1719 status
|= DO_SETLINK_NOTIFY
;
1720 dev
->link_mode
= value
;
1721 write_unlock_bh(&dev_base_lock
);
1724 if (tb
[IFLA_VFINFO_LIST
]) {
1725 struct nlattr
*attr
;
1727 nla_for_each_nested(attr
, tb
[IFLA_VFINFO_LIST
], rem
) {
1728 if (nla_type(attr
) != IFLA_VF_INFO
) {
1732 err
= do_setvfinfo(dev
, attr
);
1735 status
|= DO_SETLINK_NOTIFY
;
1740 if (tb
[IFLA_VF_PORTS
]) {
1741 struct nlattr
*port
[IFLA_PORT_MAX
+1];
1742 struct nlattr
*attr
;
1747 if (!ops
->ndo_set_vf_port
)
1750 nla_for_each_nested(attr
, tb
[IFLA_VF_PORTS
], rem
) {
1751 if (nla_type(attr
) != IFLA_VF_PORT
)
1753 err
= nla_parse_nested(port
, IFLA_PORT_MAX
,
1754 attr
, ifla_port_policy
);
1757 if (!port
[IFLA_PORT_VF
]) {
1761 vf
= nla_get_u32(port
[IFLA_PORT_VF
]);
1762 err
= ops
->ndo_set_vf_port(dev
, vf
, port
);
1765 status
|= DO_SETLINK_NOTIFY
;
1770 if (tb
[IFLA_PORT_SELF
]) {
1771 struct nlattr
*port
[IFLA_PORT_MAX
+1];
1773 err
= nla_parse_nested(port
, IFLA_PORT_MAX
,
1774 tb
[IFLA_PORT_SELF
], ifla_port_policy
);
1779 if (ops
->ndo_set_vf_port
)
1780 err
= ops
->ndo_set_vf_port(dev
, PORT_SELF_VF
, port
);
1783 status
|= DO_SETLINK_NOTIFY
;
1786 if (tb
[IFLA_AF_SPEC
]) {
1790 nla_for_each_nested(af
, tb
[IFLA_AF_SPEC
], rem
) {
1791 const struct rtnl_af_ops
*af_ops
;
1793 if (!(af_ops
= rtnl_af_lookup(nla_type(af
))))
1796 err
= af_ops
->set_link_af(dev
, af
);
1800 status
|= DO_SETLINK_NOTIFY
;
1806 if (status
& DO_SETLINK_MODIFIED
) {
1807 if (status
& DO_SETLINK_NOTIFY
)
1808 netdev_state_change(dev
);
1811 net_warn_ratelimited("A link change request failed with some changes committed already. Interface %s may have been left with an inconsistent configuration, please check.\n",
1818 static int rtnl_setlink(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
1820 struct net
*net
= sock_net(skb
->sk
);
1821 struct ifinfomsg
*ifm
;
1822 struct net_device
*dev
;
1824 struct nlattr
*tb
[IFLA_MAX
+1];
1825 char ifname
[IFNAMSIZ
];
1827 err
= nlmsg_parse(nlh
, sizeof(*ifm
), tb
, IFLA_MAX
, ifla_policy
);
1831 if (tb
[IFLA_IFNAME
])
1832 nla_strlcpy(ifname
, tb
[IFLA_IFNAME
], IFNAMSIZ
);
1837 ifm
= nlmsg_data(nlh
);
1838 if (ifm
->ifi_index
> 0)
1839 dev
= __dev_get_by_index(net
, ifm
->ifi_index
);
1840 else if (tb
[IFLA_IFNAME
])
1841 dev
= __dev_get_by_name(net
, ifname
);
1850 err
= validate_linkmsg(dev
, tb
);
1854 err
= do_setlink(skb
, dev
, ifm
, tb
, ifname
, 0);
1859 static int rtnl_group_dellink(const struct net
*net
, int group
)
1861 struct net_device
*dev
, *aux
;
1862 LIST_HEAD(list_kill
);
1868 for_each_netdev(net
, dev
) {
1869 if (dev
->group
== group
) {
1870 const struct rtnl_link_ops
*ops
;
1873 ops
= dev
->rtnl_link_ops
;
1874 if (!ops
|| !ops
->dellink
)
1882 for_each_netdev_safe(net
, dev
, aux
) {
1883 if (dev
->group
== group
) {
1884 const struct rtnl_link_ops
*ops
;
1886 ops
= dev
->rtnl_link_ops
;
1887 ops
->dellink(dev
, &list_kill
);
1890 unregister_netdevice_many(&list_kill
);
1895 static int rtnl_dellink(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
1897 struct net
*net
= sock_net(skb
->sk
);
1898 const struct rtnl_link_ops
*ops
;
1899 struct net_device
*dev
;
1900 struct ifinfomsg
*ifm
;
1901 char ifname
[IFNAMSIZ
];
1902 struct nlattr
*tb
[IFLA_MAX
+1];
1904 LIST_HEAD(list_kill
);
1906 err
= nlmsg_parse(nlh
, sizeof(*ifm
), tb
, IFLA_MAX
, ifla_policy
);
1910 if (tb
[IFLA_IFNAME
])
1911 nla_strlcpy(ifname
, tb
[IFLA_IFNAME
], IFNAMSIZ
);
1913 ifm
= nlmsg_data(nlh
);
1914 if (ifm
->ifi_index
> 0)
1915 dev
= __dev_get_by_index(net
, ifm
->ifi_index
);
1916 else if (tb
[IFLA_IFNAME
])
1917 dev
= __dev_get_by_name(net
, ifname
);
1918 else if (tb
[IFLA_GROUP
])
1919 return rtnl_group_dellink(net
, nla_get_u32(tb
[IFLA_GROUP
]));
1926 ops
= dev
->rtnl_link_ops
;
1927 if (!ops
|| !ops
->dellink
)
1930 ops
->dellink(dev
, &list_kill
);
1931 unregister_netdevice_many(&list_kill
);
1935 int rtnl_configure_link(struct net_device
*dev
, const struct ifinfomsg
*ifm
)
1937 unsigned int old_flags
;
1940 old_flags
= dev
->flags
;
1941 if (ifm
&& (ifm
->ifi_flags
|| ifm
->ifi_change
)) {
1942 err
= __dev_change_flags(dev
, rtnl_dev_combine_flags(dev
, ifm
));
1947 dev
->rtnl_link_state
= RTNL_LINK_INITIALIZED
;
1949 __dev_notify_flags(dev
, old_flags
, ~0U);
1952 EXPORT_SYMBOL(rtnl_configure_link
);
1954 struct net_device
*rtnl_create_link(struct net
*net
,
1955 const char *ifname
, unsigned char name_assign_type
,
1956 const struct rtnl_link_ops
*ops
, struct nlattr
*tb
[])
1959 struct net_device
*dev
;
1960 unsigned int num_tx_queues
= 1;
1961 unsigned int num_rx_queues
= 1;
1963 if (tb
[IFLA_NUM_TX_QUEUES
])
1964 num_tx_queues
= nla_get_u32(tb
[IFLA_NUM_TX_QUEUES
]);
1965 else if (ops
->get_num_tx_queues
)
1966 num_tx_queues
= ops
->get_num_tx_queues();
1968 if (tb
[IFLA_NUM_RX_QUEUES
])
1969 num_rx_queues
= nla_get_u32(tb
[IFLA_NUM_RX_QUEUES
]);
1970 else if (ops
->get_num_rx_queues
)
1971 num_rx_queues
= ops
->get_num_rx_queues();
1974 dev
= alloc_netdev_mqs(ops
->priv_size
, ifname
, name_assign_type
,
1975 ops
->setup
, num_tx_queues
, num_rx_queues
);
1979 dev_net_set(dev
, net
);
1980 dev
->rtnl_link_ops
= ops
;
1981 dev
->rtnl_link_state
= RTNL_LINK_INITIALIZING
;
1984 dev
->mtu
= nla_get_u32(tb
[IFLA_MTU
]);
1985 if (tb
[IFLA_ADDRESS
]) {
1986 memcpy(dev
->dev_addr
, nla_data(tb
[IFLA_ADDRESS
]),
1987 nla_len(tb
[IFLA_ADDRESS
]));
1988 dev
->addr_assign_type
= NET_ADDR_SET
;
1990 if (tb
[IFLA_BROADCAST
])
1991 memcpy(dev
->broadcast
, nla_data(tb
[IFLA_BROADCAST
]),
1992 nla_len(tb
[IFLA_BROADCAST
]));
1993 if (tb
[IFLA_TXQLEN
])
1994 dev
->tx_queue_len
= nla_get_u32(tb
[IFLA_TXQLEN
]);
1995 if (tb
[IFLA_OPERSTATE
])
1996 set_operstate(dev
, nla_get_u8(tb
[IFLA_OPERSTATE
]));
1997 if (tb
[IFLA_LINKMODE
])
1998 dev
->link_mode
= nla_get_u8(tb
[IFLA_LINKMODE
]);
2000 dev_set_group(dev
, nla_get_u32(tb
[IFLA_GROUP
]));
2005 return ERR_PTR(err
);
2007 EXPORT_SYMBOL(rtnl_create_link
);
2009 static int rtnl_group_changelink(const struct sk_buff
*skb
,
2010 struct net
*net
, int group
,
2011 struct ifinfomsg
*ifm
,
2014 struct net_device
*dev
, *aux
;
2017 for_each_netdev_safe(net
, dev
, aux
) {
2018 if (dev
->group
== group
) {
2019 err
= do_setlink(skb
, dev
, ifm
, tb
, NULL
, 0);
2028 static int rtnl_newlink(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
2030 struct net
*net
= sock_net(skb
->sk
);
2031 const struct rtnl_link_ops
*ops
;
2032 const struct rtnl_link_ops
*m_ops
= NULL
;
2033 struct net_device
*dev
;
2034 struct net_device
*master_dev
= NULL
;
2035 struct ifinfomsg
*ifm
;
2036 char kind
[MODULE_NAME_LEN
];
2037 char ifname
[IFNAMSIZ
];
2038 struct nlattr
*tb
[IFLA_MAX
+1];
2039 struct nlattr
*linkinfo
[IFLA_INFO_MAX
+1];
2040 unsigned char name_assign_type
= NET_NAME_USER
;
2043 #ifdef CONFIG_MODULES
2046 err
= nlmsg_parse(nlh
, sizeof(*ifm
), tb
, IFLA_MAX
, ifla_policy
);
2050 if (tb
[IFLA_IFNAME
])
2051 nla_strlcpy(ifname
, tb
[IFLA_IFNAME
], IFNAMSIZ
);
2055 ifm
= nlmsg_data(nlh
);
2056 if (ifm
->ifi_index
> 0)
2057 dev
= __dev_get_by_index(net
, ifm
->ifi_index
);
2060 dev
= __dev_get_by_name(net
, ifname
);
2066 master_dev
= netdev_master_upper_dev_get(dev
);
2068 m_ops
= master_dev
->rtnl_link_ops
;
2071 err
= validate_linkmsg(dev
, tb
);
2075 if (tb
[IFLA_LINKINFO
]) {
2076 err
= nla_parse_nested(linkinfo
, IFLA_INFO_MAX
,
2077 tb
[IFLA_LINKINFO
], ifla_info_policy
);
2081 memset(linkinfo
, 0, sizeof(linkinfo
));
2083 if (linkinfo
[IFLA_INFO_KIND
]) {
2084 nla_strlcpy(kind
, linkinfo
[IFLA_INFO_KIND
], sizeof(kind
));
2085 ops
= rtnl_link_ops_get(kind
);
2092 struct nlattr
*attr
[ops
? ops
->maxtype
+ 1 : 1];
2093 struct nlattr
*slave_attr
[m_ops
? m_ops
->slave_maxtype
+ 1 : 1];
2094 struct nlattr
**data
= NULL
;
2095 struct nlattr
**slave_data
= NULL
;
2096 struct net
*dest_net
, *link_net
= NULL
;
2099 if (ops
->maxtype
&& linkinfo
[IFLA_INFO_DATA
]) {
2100 err
= nla_parse_nested(attr
, ops
->maxtype
,
2101 linkinfo
[IFLA_INFO_DATA
],
2107 if (ops
->validate
) {
2108 err
= ops
->validate(tb
, data
);
2115 if (m_ops
->slave_maxtype
&&
2116 linkinfo
[IFLA_INFO_SLAVE_DATA
]) {
2117 err
= nla_parse_nested(slave_attr
,
2118 m_ops
->slave_maxtype
,
2119 linkinfo
[IFLA_INFO_SLAVE_DATA
],
2120 m_ops
->slave_policy
);
2123 slave_data
= slave_attr
;
2125 if (m_ops
->slave_validate
) {
2126 err
= m_ops
->slave_validate(tb
, slave_data
);
2135 if (nlh
->nlmsg_flags
& NLM_F_EXCL
)
2137 if (nlh
->nlmsg_flags
& NLM_F_REPLACE
)
2140 if (linkinfo
[IFLA_INFO_DATA
]) {
2141 if (!ops
|| ops
!= dev
->rtnl_link_ops
||
2145 err
= ops
->changelink(dev
, tb
, data
);
2148 status
|= DO_SETLINK_NOTIFY
;
2151 if (linkinfo
[IFLA_INFO_SLAVE_DATA
]) {
2152 if (!m_ops
|| !m_ops
->slave_changelink
)
2155 err
= m_ops
->slave_changelink(master_dev
, dev
,
2159 status
|= DO_SETLINK_NOTIFY
;
2162 return do_setlink(skb
, dev
, ifm
, tb
, ifname
, status
);
2165 if (!(nlh
->nlmsg_flags
& NLM_F_CREATE
)) {
2166 if (ifm
->ifi_index
== 0 && tb
[IFLA_GROUP
])
2167 return rtnl_group_changelink(skb
, net
,
2168 nla_get_u32(tb
[IFLA_GROUP
]),
2173 if (tb
[IFLA_MAP
] || tb
[IFLA_MASTER
] || tb
[IFLA_PROTINFO
])
2177 #ifdef CONFIG_MODULES
2180 request_module("rtnl-link-%s", kind
);
2182 ops
= rtnl_link_ops_get(kind
);
2194 snprintf(ifname
, IFNAMSIZ
, "%s%%d", ops
->kind
);
2195 name_assign_type
= NET_NAME_ENUM
;
2198 dest_net
= rtnl_link_get_net(net
, tb
);
2199 if (IS_ERR(dest_net
))
2200 return PTR_ERR(dest_net
);
2203 if (!netlink_ns_capable(skb
, dest_net
->user_ns
, CAP_NET_ADMIN
))
2206 if (tb
[IFLA_LINK_NETNSID
]) {
2207 int id
= nla_get_s32(tb
[IFLA_LINK_NETNSID
]);
2209 link_net
= get_net_ns_by_id(dest_net
, id
);
2215 if (!netlink_ns_capable(skb
, link_net
->user_ns
, CAP_NET_ADMIN
))
2219 dev
= rtnl_create_link(link_net
? : dest_net
, ifname
,
2220 name_assign_type
, ops
, tb
);
2226 dev
->ifindex
= ifm
->ifi_index
;
2229 err
= ops
->newlink(link_net
? : net
, dev
, tb
, data
);
2230 /* Drivers should call free_netdev() in ->destructor
2231 * and unregister it on failure after registration
2232 * so that device could be finally freed in rtnl_unlock.
2235 /* If device is not registered at all, free it now */
2236 if (dev
->reg_state
== NETREG_UNINITIALIZED
)
2241 err
= register_netdevice(dev
);
2247 err
= rtnl_configure_link(dev
, ifm
);
2249 goto out_unregister
;
2251 err
= dev_change_net_namespace(dev
, dest_net
, ifname
);
2253 goto out_unregister
;
2262 LIST_HEAD(list_kill
);
2264 ops
->dellink(dev
, &list_kill
);
2265 unregister_netdevice_many(&list_kill
);
2267 unregister_netdevice(dev
);
2273 static int rtnl_getlink(struct sk_buff
*skb
, struct nlmsghdr
* nlh
)
2275 struct net
*net
= sock_net(skb
->sk
);
2276 struct ifinfomsg
*ifm
;
2277 char ifname
[IFNAMSIZ
];
2278 struct nlattr
*tb
[IFLA_MAX
+1];
2279 struct net_device
*dev
= NULL
;
2280 struct sk_buff
*nskb
;
2282 u32 ext_filter_mask
= 0;
2284 err
= nlmsg_parse(nlh
, sizeof(*ifm
), tb
, IFLA_MAX
, ifla_policy
);
2288 if (tb
[IFLA_IFNAME
])
2289 nla_strlcpy(ifname
, tb
[IFLA_IFNAME
], IFNAMSIZ
);
2291 if (tb
[IFLA_EXT_MASK
])
2292 ext_filter_mask
= nla_get_u32(tb
[IFLA_EXT_MASK
]);
2294 ifm
= nlmsg_data(nlh
);
2295 if (ifm
->ifi_index
> 0)
2296 dev
= __dev_get_by_index(net
, ifm
->ifi_index
);
2297 else if (tb
[IFLA_IFNAME
])
2298 dev
= __dev_get_by_name(net
, ifname
);
2305 nskb
= nlmsg_new(if_nlmsg_size(dev
, ext_filter_mask
), GFP_KERNEL
);
2309 err
= rtnl_fill_ifinfo(nskb
, dev
, RTM_NEWLINK
, NETLINK_CB(skb
).portid
,
2310 nlh
->nlmsg_seq
, 0, 0, ext_filter_mask
);
2312 /* -EMSGSIZE implies BUG in if_nlmsg_size */
2313 WARN_ON(err
== -EMSGSIZE
);
2316 err
= rtnl_unicast(nskb
, net
, NETLINK_CB(skb
).portid
);
2321 static u16
rtnl_calcit(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
2323 struct net
*net
= sock_net(skb
->sk
);
2324 struct net_device
*dev
;
2325 struct nlattr
*tb
[IFLA_MAX
+1];
2326 u32 ext_filter_mask
= 0;
2327 u16 min_ifinfo_dump_size
= 0;
2330 /* Same kernel<->userspace interface hack as in rtnl_dump_ifinfo. */
2331 hdrlen
= nlmsg_len(nlh
) < sizeof(struct ifinfomsg
) ?
2332 sizeof(struct rtgenmsg
) : sizeof(struct ifinfomsg
);
2334 if (nlmsg_parse(nlh
, hdrlen
, tb
, IFLA_MAX
, ifla_policy
) >= 0) {
2335 if (tb
[IFLA_EXT_MASK
])
2336 ext_filter_mask
= nla_get_u32(tb
[IFLA_EXT_MASK
]);
2339 if (!ext_filter_mask
)
2340 return NLMSG_GOODSIZE
;
2342 * traverse the list of net devices and compute the minimum
2343 * buffer size based upon the filter mask.
2345 list_for_each_entry(dev
, &net
->dev_base_head
, dev_list
) {
2346 min_ifinfo_dump_size
= max_t(u16
, min_ifinfo_dump_size
,
2351 return min_ifinfo_dump_size
;
2354 static int rtnl_dump_all(struct sk_buff
*skb
, struct netlink_callback
*cb
)
2357 int s_idx
= cb
->family
;
2361 for (idx
= 1; idx
<= RTNL_FAMILY_MAX
; idx
++) {
2362 int type
= cb
->nlh
->nlmsg_type
-RTM_BASE
;
2363 if (idx
< s_idx
|| idx
== PF_PACKET
)
2365 if (rtnl_msg_handlers
[idx
] == NULL
||
2366 rtnl_msg_handlers
[idx
][type
].dumpit
== NULL
)
2369 memset(&cb
->args
[0], 0, sizeof(cb
->args
));
2373 if (rtnl_msg_handlers
[idx
][type
].dumpit(skb
, cb
))
2381 struct sk_buff
*rtmsg_ifinfo_build_skb(int type
, struct net_device
*dev
,
2382 unsigned int change
, gfp_t flags
)
2384 struct net
*net
= dev_net(dev
);
2385 struct sk_buff
*skb
;
2387 size_t if_info_size
;
2389 skb
= nlmsg_new((if_info_size
= if_nlmsg_size(dev
, 0)), flags
);
2393 err
= rtnl_fill_ifinfo(skb
, dev
, type
, 0, 0, change
, 0, 0);
2395 /* -EMSGSIZE implies BUG in if_nlmsg_size() */
2396 WARN_ON(err
== -EMSGSIZE
);
2403 rtnl_set_sk_err(net
, RTNLGRP_LINK
, err
);
2407 void rtmsg_ifinfo_send(struct sk_buff
*skb
, struct net_device
*dev
, gfp_t flags
)
2409 struct net
*net
= dev_net(dev
);
2411 rtnl_notify(skb
, net
, 0, RTNLGRP_LINK
, NULL
, flags
);
2414 void rtmsg_ifinfo(int type
, struct net_device
*dev
, unsigned int change
,
2417 struct sk_buff
*skb
;
2419 skb
= rtmsg_ifinfo_build_skb(type
, dev
, change
, flags
);
2421 rtmsg_ifinfo_send(skb
, dev
, flags
);
2423 EXPORT_SYMBOL(rtmsg_ifinfo
);
2425 static int nlmsg_populate_fdb_fill(struct sk_buff
*skb
,
2426 struct net_device
*dev
,
2427 u8
*addr
, u16 vid
, u32 pid
, u32 seq
,
2428 int type
, unsigned int flags
,
2431 struct nlmsghdr
*nlh
;
2434 nlh
= nlmsg_put(skb
, pid
, seq
, type
, sizeof(*ndm
), nlflags
);
2438 ndm
= nlmsg_data(nlh
);
2439 ndm
->ndm_family
= AF_BRIDGE
;
2442 ndm
->ndm_flags
= flags
;
2444 ndm
->ndm_ifindex
= dev
->ifindex
;
2445 ndm
->ndm_state
= NUD_PERMANENT
;
2447 if (nla_put(skb
, NDA_LLADDR
, ETH_ALEN
, addr
))
2448 goto nla_put_failure
;
2450 if (nla_put(skb
, NDA_VLAN
, sizeof(u16
), &vid
))
2451 goto nla_put_failure
;
2453 nlmsg_end(skb
, nlh
);
2457 nlmsg_cancel(skb
, nlh
);
2461 static inline size_t rtnl_fdb_nlmsg_size(void)
2463 return NLMSG_ALIGN(sizeof(struct ndmsg
)) + nla_total_size(ETH_ALEN
);
2466 static void rtnl_fdb_notify(struct net_device
*dev
, u8
*addr
, u16 vid
, int type
)
2468 struct net
*net
= dev_net(dev
);
2469 struct sk_buff
*skb
;
2472 skb
= nlmsg_new(rtnl_fdb_nlmsg_size(), GFP_ATOMIC
);
2476 err
= nlmsg_populate_fdb_fill(skb
, dev
, addr
, vid
,
2477 0, 0, type
, NTF_SELF
, 0);
2483 rtnl_notify(skb
, net
, 0, RTNLGRP_NEIGH
, NULL
, GFP_ATOMIC
);
2486 rtnl_set_sk_err(net
, RTNLGRP_NEIGH
, err
);
2490 * ndo_dflt_fdb_add - default netdevice operation to add an FDB entry
2492 int ndo_dflt_fdb_add(struct ndmsg
*ndm
,
2493 struct nlattr
*tb
[],
2494 struct net_device
*dev
,
2495 const unsigned char *addr
, u16 vid
,
2500 /* If aging addresses are supported device will need to
2501 * implement its own handler for this.
2503 if (ndm
->ndm_state
&& !(ndm
->ndm_state
& NUD_PERMANENT
)) {
2504 pr_info("%s: FDB only supports static addresses\n", dev
->name
);
2509 pr_info("%s: vlans aren't supported yet for dev_uc|mc_add()\n", dev
->name
);
2513 if (is_unicast_ether_addr(addr
) || is_link_local_ether_addr(addr
))
2514 err
= dev_uc_add_excl(dev
, addr
);
2515 else if (is_multicast_ether_addr(addr
))
2516 err
= dev_mc_add_excl(dev
, addr
);
2518 /* Only return duplicate errors if NLM_F_EXCL is set */
2519 if (err
== -EEXIST
&& !(flags
& NLM_F_EXCL
))
2524 EXPORT_SYMBOL(ndo_dflt_fdb_add
);
2526 static int fdb_vid_parse(struct nlattr
*vlan_attr
, u16
*p_vid
)
2531 if (nla_len(vlan_attr
) != sizeof(u16
)) {
2532 pr_info("PF_BRIDGE: RTM_NEWNEIGH with invalid vlan\n");
2536 vid
= nla_get_u16(vlan_attr
);
2538 if (!vid
|| vid
>= VLAN_VID_MASK
) {
2539 pr_info("PF_BRIDGE: RTM_NEWNEIGH with invalid vlan id %d\n",
2548 static int rtnl_fdb_add(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
2550 struct net
*net
= sock_net(skb
->sk
);
2552 struct nlattr
*tb
[NDA_MAX
+1];
2553 struct net_device
*dev
;
2558 err
= nlmsg_parse(nlh
, sizeof(*ndm
), tb
, NDA_MAX
, NULL
);
2562 ndm
= nlmsg_data(nlh
);
2563 if (ndm
->ndm_ifindex
== 0) {
2564 pr_info("PF_BRIDGE: RTM_NEWNEIGH with invalid ifindex\n");
2568 dev
= __dev_get_by_index(net
, ndm
->ndm_ifindex
);
2570 pr_info("PF_BRIDGE: RTM_NEWNEIGH with unknown ifindex\n");
2574 if (!tb
[NDA_LLADDR
] || nla_len(tb
[NDA_LLADDR
]) != ETH_ALEN
) {
2575 pr_info("PF_BRIDGE: RTM_NEWNEIGH with invalid address\n");
2579 addr
= nla_data(tb
[NDA_LLADDR
]);
2581 err
= fdb_vid_parse(tb
[NDA_VLAN
], &vid
);
2587 /* Support fdb on master device the net/bridge default case */
2588 if ((!ndm
->ndm_flags
|| ndm
->ndm_flags
& NTF_MASTER
) &&
2589 (dev
->priv_flags
& IFF_BRIDGE_PORT
)) {
2590 struct net_device
*br_dev
= netdev_master_upper_dev_get(dev
);
2591 const struct net_device_ops
*ops
= br_dev
->netdev_ops
;
2593 err
= ops
->ndo_fdb_add(ndm
, tb
, dev
, addr
, vid
,
2598 ndm
->ndm_flags
&= ~NTF_MASTER
;
2601 /* Embedded bridge, macvlan, and any other device support */
2602 if ((ndm
->ndm_flags
& NTF_SELF
)) {
2603 if (dev
->netdev_ops
->ndo_fdb_add
)
2604 err
= dev
->netdev_ops
->ndo_fdb_add(ndm
, tb
, dev
, addr
,
2608 err
= ndo_dflt_fdb_add(ndm
, tb
, dev
, addr
, vid
,
2612 rtnl_fdb_notify(dev
, addr
, vid
, RTM_NEWNEIGH
);
2613 ndm
->ndm_flags
&= ~NTF_SELF
;
2621 * ndo_dflt_fdb_del - default netdevice operation to delete an FDB entry
2623 int ndo_dflt_fdb_del(struct ndmsg
*ndm
,
2624 struct nlattr
*tb
[],
2625 struct net_device
*dev
,
2626 const unsigned char *addr
, u16 vid
)
2630 /* If aging addresses are supported device will need to
2631 * implement its own handler for this.
2633 if (!(ndm
->ndm_state
& NUD_PERMANENT
)) {
2634 pr_info("%s: FDB only supports static addresses\n", dev
->name
);
2638 if (is_unicast_ether_addr(addr
) || is_link_local_ether_addr(addr
))
2639 err
= dev_uc_del(dev
, addr
);
2640 else if (is_multicast_ether_addr(addr
))
2641 err
= dev_mc_del(dev
, addr
);
2645 EXPORT_SYMBOL(ndo_dflt_fdb_del
);
2647 static int rtnl_fdb_del(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
2649 struct net
*net
= sock_net(skb
->sk
);
2651 struct nlattr
*tb
[NDA_MAX
+1];
2652 struct net_device
*dev
;
2657 if (!netlink_capable(skb
, CAP_NET_ADMIN
))
2660 err
= nlmsg_parse(nlh
, sizeof(*ndm
), tb
, NDA_MAX
, NULL
);
2664 ndm
= nlmsg_data(nlh
);
2665 if (ndm
->ndm_ifindex
== 0) {
2666 pr_info("PF_BRIDGE: RTM_DELNEIGH with invalid ifindex\n");
2670 dev
= __dev_get_by_index(net
, ndm
->ndm_ifindex
);
2672 pr_info("PF_BRIDGE: RTM_DELNEIGH with unknown ifindex\n");
2676 if (!tb
[NDA_LLADDR
] || nla_len(tb
[NDA_LLADDR
]) != ETH_ALEN
) {
2677 pr_info("PF_BRIDGE: RTM_DELNEIGH with invalid address\n");
2681 addr
= nla_data(tb
[NDA_LLADDR
]);
2683 err
= fdb_vid_parse(tb
[NDA_VLAN
], &vid
);
2689 /* Support fdb on master device the net/bridge default case */
2690 if ((!ndm
->ndm_flags
|| ndm
->ndm_flags
& NTF_MASTER
) &&
2691 (dev
->priv_flags
& IFF_BRIDGE_PORT
)) {
2692 struct net_device
*br_dev
= netdev_master_upper_dev_get(dev
);
2693 const struct net_device_ops
*ops
= br_dev
->netdev_ops
;
2695 if (ops
->ndo_fdb_del
)
2696 err
= ops
->ndo_fdb_del(ndm
, tb
, dev
, addr
, vid
);
2701 ndm
->ndm_flags
&= ~NTF_MASTER
;
2704 /* Embedded bridge, macvlan, and any other device support */
2705 if (ndm
->ndm_flags
& NTF_SELF
) {
2706 if (dev
->netdev_ops
->ndo_fdb_del
)
2707 err
= dev
->netdev_ops
->ndo_fdb_del(ndm
, tb
, dev
, addr
,
2710 err
= ndo_dflt_fdb_del(ndm
, tb
, dev
, addr
, vid
);
2713 rtnl_fdb_notify(dev
, addr
, vid
, RTM_DELNEIGH
);
2714 ndm
->ndm_flags
&= ~NTF_SELF
;
2721 static int nlmsg_populate_fdb(struct sk_buff
*skb
,
2722 struct netlink_callback
*cb
,
2723 struct net_device
*dev
,
2725 struct netdev_hw_addr_list
*list
)
2727 struct netdev_hw_addr
*ha
;
2731 portid
= NETLINK_CB(cb
->skb
).portid
;
2732 seq
= cb
->nlh
->nlmsg_seq
;
2734 list_for_each_entry(ha
, &list
->list
, list
) {
2735 if (*idx
< cb
->args
[0])
2738 err
= nlmsg_populate_fdb_fill(skb
, dev
, ha
->addr
, 0,
2740 RTM_NEWNEIGH
, NTF_SELF
,
2751 * ndo_dflt_fdb_dump - default netdevice operation to dump an FDB table.
2752 * @nlh: netlink message header
2755 * Default netdevice operation to dump the existing unicast address list.
2756 * Returns number of addresses from list put in skb.
2758 int ndo_dflt_fdb_dump(struct sk_buff
*skb
,
2759 struct netlink_callback
*cb
,
2760 struct net_device
*dev
,
2761 struct net_device
*filter_dev
,
2766 netif_addr_lock_bh(dev
);
2767 err
= nlmsg_populate_fdb(skb
, cb
, dev
, &idx
, &dev
->uc
);
2770 nlmsg_populate_fdb(skb
, cb
, dev
, &idx
, &dev
->mc
);
2772 netif_addr_unlock_bh(dev
);
2775 EXPORT_SYMBOL(ndo_dflt_fdb_dump
);
2777 static int rtnl_fdb_dump(struct sk_buff
*skb
, struct netlink_callback
*cb
)
2779 struct net_device
*dev
;
2780 struct nlattr
*tb
[IFLA_MAX
+1];
2781 struct net_device
*br_dev
= NULL
;
2782 const struct net_device_ops
*ops
= NULL
;
2783 const struct net_device_ops
*cops
= NULL
;
2784 struct ifinfomsg
*ifm
= nlmsg_data(cb
->nlh
);
2785 struct net
*net
= sock_net(skb
->sk
);
2790 if (nlmsg_parse(cb
->nlh
, sizeof(struct ifinfomsg
), tb
, IFLA_MAX
,
2791 ifla_policy
) == 0) {
2792 if (tb
[IFLA_MASTER
])
2793 br_idx
= nla_get_u32(tb
[IFLA_MASTER
]);
2796 brport_idx
= ifm
->ifi_index
;
2799 br_dev
= __dev_get_by_index(net
, br_idx
);
2803 ops
= br_dev
->netdev_ops
;
2806 for_each_netdev(net
, dev
) {
2807 if (brport_idx
&& (dev
->ifindex
!= brport_idx
))
2810 if (!br_idx
) { /* user did not specify a specific bridge */
2811 if (dev
->priv_flags
& IFF_BRIDGE_PORT
) {
2812 br_dev
= netdev_master_upper_dev_get(dev
);
2813 cops
= br_dev
->netdev_ops
;
2817 if (dev
!= br_dev
&&
2818 !(dev
->priv_flags
& IFF_BRIDGE_PORT
))
2821 if (br_dev
!= netdev_master_upper_dev_get(dev
) &&
2822 !(dev
->priv_flags
& IFF_EBRIDGE
))
2828 if (dev
->priv_flags
& IFF_BRIDGE_PORT
) {
2829 if (cops
&& cops
->ndo_fdb_dump
)
2830 idx
= cops
->ndo_fdb_dump(skb
, cb
, br_dev
, dev
,
2834 if (dev
->netdev_ops
->ndo_fdb_dump
)
2835 idx
= dev
->netdev_ops
->ndo_fdb_dump(skb
, cb
, dev
, NULL
,
2838 idx
= ndo_dflt_fdb_dump(skb
, cb
, dev
, NULL
, idx
);
2847 static int brport_nla_put_flag(struct sk_buff
*skb
, u32 flags
, u32 mask
,
2848 unsigned int attrnum
, unsigned int flag
)
2851 return nla_put_u8(skb
, attrnum
, !!(flags
& flag
));
2855 int ndo_dflt_bridge_getlink(struct sk_buff
*skb
, u32 pid
, u32 seq
,
2856 struct net_device
*dev
, u16 mode
,
2857 u32 flags
, u32 mask
)
2859 struct nlmsghdr
*nlh
;
2860 struct ifinfomsg
*ifm
;
2861 struct nlattr
*br_afspec
;
2862 struct nlattr
*protinfo
;
2863 u8 operstate
= netif_running(dev
) ? dev
->operstate
: IF_OPER_DOWN
;
2864 struct net_device
*br_dev
= netdev_master_upper_dev_get(dev
);
2866 nlh
= nlmsg_put(skb
, pid
, seq
, RTM_NEWLINK
, sizeof(*ifm
), NLM_F_MULTI
);
2870 ifm
= nlmsg_data(nlh
);
2871 ifm
->ifi_family
= AF_BRIDGE
;
2873 ifm
->ifi_type
= dev
->type
;
2874 ifm
->ifi_index
= dev
->ifindex
;
2875 ifm
->ifi_flags
= dev_get_flags(dev
);
2876 ifm
->ifi_change
= 0;
2879 if (nla_put_string(skb
, IFLA_IFNAME
, dev
->name
) ||
2880 nla_put_u32(skb
, IFLA_MTU
, dev
->mtu
) ||
2881 nla_put_u8(skb
, IFLA_OPERSTATE
, operstate
) ||
2883 nla_put_u32(skb
, IFLA_MASTER
, br_dev
->ifindex
)) ||
2885 nla_put(skb
, IFLA_ADDRESS
, dev
->addr_len
, dev
->dev_addr
)) ||
2886 (dev
->ifindex
!= dev_get_iflink(dev
) &&
2887 nla_put_u32(skb
, IFLA_LINK
, dev_get_iflink(dev
))))
2888 goto nla_put_failure
;
2890 br_afspec
= nla_nest_start(skb
, IFLA_AF_SPEC
);
2892 goto nla_put_failure
;
2894 if (nla_put_u16(skb
, IFLA_BRIDGE_FLAGS
, BRIDGE_FLAGS_SELF
)) {
2895 nla_nest_cancel(skb
, br_afspec
);
2896 goto nla_put_failure
;
2899 if (mode
!= BRIDGE_MODE_UNDEF
) {
2900 if (nla_put_u16(skb
, IFLA_BRIDGE_MODE
, mode
)) {
2901 nla_nest_cancel(skb
, br_afspec
);
2902 goto nla_put_failure
;
2905 nla_nest_end(skb
, br_afspec
);
2907 protinfo
= nla_nest_start(skb
, IFLA_PROTINFO
| NLA_F_NESTED
);
2909 goto nla_put_failure
;
2911 if (brport_nla_put_flag(skb
, flags
, mask
,
2912 IFLA_BRPORT_MODE
, BR_HAIRPIN_MODE
) ||
2913 brport_nla_put_flag(skb
, flags
, mask
,
2914 IFLA_BRPORT_GUARD
, BR_BPDU_GUARD
) ||
2915 brport_nla_put_flag(skb
, flags
, mask
,
2916 IFLA_BRPORT_FAST_LEAVE
,
2917 BR_MULTICAST_FAST_LEAVE
) ||
2918 brport_nla_put_flag(skb
, flags
, mask
,
2919 IFLA_BRPORT_PROTECT
, BR_ROOT_BLOCK
) ||
2920 brport_nla_put_flag(skb
, flags
, mask
,
2921 IFLA_BRPORT_LEARNING
, BR_LEARNING
) ||
2922 brport_nla_put_flag(skb
, flags
, mask
,
2923 IFLA_BRPORT_LEARNING_SYNC
, BR_LEARNING_SYNC
) ||
2924 brport_nla_put_flag(skb
, flags
, mask
,
2925 IFLA_BRPORT_UNICAST_FLOOD
, BR_FLOOD
) ||
2926 brport_nla_put_flag(skb
, flags
, mask
,
2927 IFLA_BRPORT_PROXYARP
, BR_PROXYARP
)) {
2928 nla_nest_cancel(skb
, protinfo
);
2929 goto nla_put_failure
;
2932 nla_nest_end(skb
, protinfo
);
2934 nlmsg_end(skb
, nlh
);
2937 nlmsg_cancel(skb
, nlh
);
2940 EXPORT_SYMBOL(ndo_dflt_bridge_getlink
);
2942 static int rtnl_bridge_getlink(struct sk_buff
*skb
, struct netlink_callback
*cb
)
2944 struct net
*net
= sock_net(skb
->sk
);
2945 struct net_device
*dev
;
2947 u32 portid
= NETLINK_CB(cb
->skb
).portid
;
2948 u32 seq
= cb
->nlh
->nlmsg_seq
;
2949 u32 filter_mask
= 0;
2951 if (nlmsg_len(cb
->nlh
) > sizeof(struct ifinfomsg
)) {
2952 struct nlattr
*extfilt
;
2954 extfilt
= nlmsg_find_attr(cb
->nlh
, sizeof(struct ifinfomsg
),
2957 if (nla_len(extfilt
) < sizeof(filter_mask
))
2960 filter_mask
= nla_get_u32(extfilt
);
2965 for_each_netdev_rcu(net
, dev
) {
2966 const struct net_device_ops
*ops
= dev
->netdev_ops
;
2967 struct net_device
*br_dev
= netdev_master_upper_dev_get(dev
);
2969 if (br_dev
&& br_dev
->netdev_ops
->ndo_bridge_getlink
) {
2970 if (idx
>= cb
->args
[0] &&
2971 br_dev
->netdev_ops
->ndo_bridge_getlink(
2972 skb
, portid
, seq
, dev
, filter_mask
) < 0)
2977 if (ops
->ndo_bridge_getlink
) {
2978 if (idx
>= cb
->args
[0] &&
2979 ops
->ndo_bridge_getlink(skb
, portid
, seq
, dev
,
2991 static inline size_t bridge_nlmsg_size(void)
2993 return NLMSG_ALIGN(sizeof(struct ifinfomsg
))
2994 + nla_total_size(IFNAMSIZ
) /* IFLA_IFNAME */
2995 + nla_total_size(MAX_ADDR_LEN
) /* IFLA_ADDRESS */
2996 + nla_total_size(sizeof(u32
)) /* IFLA_MASTER */
2997 + nla_total_size(sizeof(u32
)) /* IFLA_MTU */
2998 + nla_total_size(sizeof(u32
)) /* IFLA_LINK */
2999 + nla_total_size(sizeof(u32
)) /* IFLA_OPERSTATE */
3000 + nla_total_size(sizeof(u8
)) /* IFLA_PROTINFO */
3001 + nla_total_size(sizeof(struct nlattr
)) /* IFLA_AF_SPEC */
3002 + nla_total_size(sizeof(u16
)) /* IFLA_BRIDGE_FLAGS */
3003 + nla_total_size(sizeof(u16
)); /* IFLA_BRIDGE_MODE */
3006 static int rtnl_bridge_notify(struct net_device
*dev
)
3008 struct net
*net
= dev_net(dev
);
3009 struct sk_buff
*skb
;
3010 int err
= -EOPNOTSUPP
;
3012 if (!dev
->netdev_ops
->ndo_bridge_getlink
)
3015 skb
= nlmsg_new(bridge_nlmsg_size(), GFP_ATOMIC
);
3021 err
= dev
->netdev_ops
->ndo_bridge_getlink(skb
, 0, 0, dev
, 0);
3028 rtnl_notify(skb
, net
, 0, RTNLGRP_LINK
, NULL
, GFP_ATOMIC
);
3031 WARN_ON(err
== -EMSGSIZE
);
3034 rtnl_set_sk_err(net
, RTNLGRP_LINK
, err
);
3038 static int rtnl_bridge_setlink(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
3040 struct net
*net
= sock_net(skb
->sk
);
3041 struct ifinfomsg
*ifm
;
3042 struct net_device
*dev
;
3043 struct nlattr
*br_spec
, *attr
= NULL
;
3044 int rem
, err
= -EOPNOTSUPP
;
3046 bool have_flags
= false;
3048 if (nlmsg_len(nlh
) < sizeof(*ifm
))
3051 ifm
= nlmsg_data(nlh
);
3052 if (ifm
->ifi_family
!= AF_BRIDGE
)
3053 return -EPFNOSUPPORT
;
3055 dev
= __dev_get_by_index(net
, ifm
->ifi_index
);
3057 pr_info("PF_BRIDGE: RTM_SETLINK with unknown ifindex\n");
3061 br_spec
= nlmsg_find_attr(nlh
, sizeof(struct ifinfomsg
), IFLA_AF_SPEC
);
3063 nla_for_each_nested(attr
, br_spec
, rem
) {
3064 if (nla_type(attr
) == IFLA_BRIDGE_FLAGS
) {
3065 if (nla_len(attr
) < sizeof(flags
))
3069 flags
= nla_get_u16(attr
);
3075 if (!flags
|| (flags
& BRIDGE_FLAGS_MASTER
)) {
3076 struct net_device
*br_dev
= netdev_master_upper_dev_get(dev
);
3078 if (!br_dev
|| !br_dev
->netdev_ops
->ndo_bridge_setlink
) {
3083 err
= br_dev
->netdev_ops
->ndo_bridge_setlink(dev
, nlh
, flags
);
3087 flags
&= ~BRIDGE_FLAGS_MASTER
;
3090 if ((flags
& BRIDGE_FLAGS_SELF
)) {
3091 if (!dev
->netdev_ops
->ndo_bridge_setlink
)
3094 err
= dev
->netdev_ops
->ndo_bridge_setlink(dev
, nlh
,
3097 flags
&= ~BRIDGE_FLAGS_SELF
;
3099 /* Generate event to notify upper layer of bridge
3102 err
= rtnl_bridge_notify(dev
);
3107 memcpy(nla_data(attr
), &flags
, sizeof(flags
));
3112 static int rtnl_bridge_dellink(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
3114 struct net
*net
= sock_net(skb
->sk
);
3115 struct ifinfomsg
*ifm
;
3116 struct net_device
*dev
;
3117 struct nlattr
*br_spec
, *attr
= NULL
;
3118 int rem
, err
= -EOPNOTSUPP
;
3120 bool have_flags
= false;
3122 if (nlmsg_len(nlh
) < sizeof(*ifm
))
3125 ifm
= nlmsg_data(nlh
);
3126 if (ifm
->ifi_family
!= AF_BRIDGE
)
3127 return -EPFNOSUPPORT
;
3129 dev
= __dev_get_by_index(net
, ifm
->ifi_index
);
3131 pr_info("PF_BRIDGE: RTM_SETLINK with unknown ifindex\n");
3135 br_spec
= nlmsg_find_attr(nlh
, sizeof(struct ifinfomsg
), IFLA_AF_SPEC
);
3137 nla_for_each_nested(attr
, br_spec
, rem
) {
3138 if (nla_type(attr
) == IFLA_BRIDGE_FLAGS
) {
3139 if (nla_len(attr
) < sizeof(flags
))
3143 flags
= nla_get_u16(attr
);
3149 if (!flags
|| (flags
& BRIDGE_FLAGS_MASTER
)) {
3150 struct net_device
*br_dev
= netdev_master_upper_dev_get(dev
);
3152 if (!br_dev
|| !br_dev
->netdev_ops
->ndo_bridge_dellink
) {
3157 err
= br_dev
->netdev_ops
->ndo_bridge_dellink(dev
, nlh
, flags
);
3161 flags
&= ~BRIDGE_FLAGS_MASTER
;
3164 if ((flags
& BRIDGE_FLAGS_SELF
)) {
3165 if (!dev
->netdev_ops
->ndo_bridge_dellink
)
3168 err
= dev
->netdev_ops
->ndo_bridge_dellink(dev
, nlh
,
3172 flags
&= ~BRIDGE_FLAGS_SELF
;
3174 /* Generate event to notify upper layer of bridge
3177 err
= rtnl_bridge_notify(dev
);
3182 memcpy(nla_data(attr
), &flags
, sizeof(flags
));
3187 /* Process one rtnetlink message. */
3189 static int rtnetlink_rcv_msg(struct sk_buff
*skb
, struct nlmsghdr
*nlh
)
3191 struct net
*net
= sock_net(skb
->sk
);
3192 rtnl_doit_func doit
;
3198 type
= nlh
->nlmsg_type
;
3204 /* All the messages must have at least 1 byte length */
3205 if (nlmsg_len(nlh
) < sizeof(struct rtgenmsg
))
3208 family
= ((struct rtgenmsg
*)nlmsg_data(nlh
))->rtgen_family
;
3212 if (kind
!= 2 && !netlink_net_capable(skb
, CAP_NET_ADMIN
))
3215 if (kind
== 2 && nlh
->nlmsg_flags
&NLM_F_DUMP
) {
3217 rtnl_dumpit_func dumpit
;
3218 rtnl_calcit_func calcit
;
3219 u16 min_dump_alloc
= 0;
3221 dumpit
= rtnl_get_dumpit(family
, type
);
3224 calcit
= rtnl_get_calcit(family
, type
);
3226 min_dump_alloc
= calcit(skb
, nlh
);
3231 struct netlink_dump_control c
= {
3233 .min_dump_alloc
= min_dump_alloc
,
3235 err
= netlink_dump_start(rtnl
, skb
, nlh
, &c
);
3241 doit
= rtnl_get_doit(family
, type
);
3245 return doit(skb
, nlh
);
3248 static void rtnetlink_rcv(struct sk_buff
*skb
)
3251 netlink_rcv_skb(skb
, &rtnetlink_rcv_msg
);
3255 static int rtnetlink_event(struct notifier_block
*this, unsigned long event
, void *ptr
)
3257 struct net_device
*dev
= netdev_notifier_info_to_dev(ptr
);
3263 case NETDEV_POST_INIT
:
3264 case NETDEV_REGISTER
:
3266 case NETDEV_PRE_TYPE_CHANGE
:
3267 case NETDEV_GOING_DOWN
:
3268 case NETDEV_UNREGISTER
:
3269 case NETDEV_UNREGISTER_FINAL
:
3270 case NETDEV_RELEASE
:
3272 case NETDEV_BONDING_INFO
:
3275 rtmsg_ifinfo(RTM_NEWLINK
, dev
, 0, GFP_KERNEL
);
3281 static struct notifier_block rtnetlink_dev_notifier
= {
3282 .notifier_call
= rtnetlink_event
,
3286 static int __net_init
rtnetlink_net_init(struct net
*net
)
3289 struct netlink_kernel_cfg cfg
= {
3290 .groups
= RTNLGRP_MAX
,
3291 .input
= rtnetlink_rcv
,
3292 .cb_mutex
= &rtnl_mutex
,
3293 .flags
= NL_CFG_F_NONROOT_RECV
,
3296 sk
= netlink_kernel_create(net
, NETLINK_ROUTE
, &cfg
);
3303 static void __net_exit
rtnetlink_net_exit(struct net
*net
)
3305 netlink_kernel_release(net
->rtnl
);
3309 static struct pernet_operations rtnetlink_net_ops
= {
3310 .init
= rtnetlink_net_init
,
3311 .exit
= rtnetlink_net_exit
,
3314 void __init
rtnetlink_init(void)
3316 if (register_pernet_subsys(&rtnetlink_net_ops
))
3317 panic("rtnetlink_init: cannot initialize rtnetlink\n");
3319 register_netdevice_notifier(&rtnetlink_dev_notifier
);
3321 rtnl_register(PF_UNSPEC
, RTM_GETLINK
, rtnl_getlink
,
3322 rtnl_dump_ifinfo
, rtnl_calcit
);
3323 rtnl_register(PF_UNSPEC
, RTM_SETLINK
, rtnl_setlink
, NULL
, NULL
);
3324 rtnl_register(PF_UNSPEC
, RTM_NEWLINK
, rtnl_newlink
, NULL
, NULL
);
3325 rtnl_register(PF_UNSPEC
, RTM_DELLINK
, rtnl_dellink
, NULL
, NULL
);
3327 rtnl_register(PF_UNSPEC
, RTM_GETADDR
, NULL
, rtnl_dump_all
, NULL
);
3328 rtnl_register(PF_UNSPEC
, RTM_GETROUTE
, NULL
, rtnl_dump_all
, NULL
);
3330 rtnl_register(PF_BRIDGE
, RTM_NEWNEIGH
, rtnl_fdb_add
, NULL
, NULL
);
3331 rtnl_register(PF_BRIDGE
, RTM_DELNEIGH
, rtnl_fdb_del
, NULL
, NULL
);
3332 rtnl_register(PF_BRIDGE
, RTM_GETNEIGH
, NULL
, rtnl_fdb_dump
, NULL
);
3334 rtnl_register(PF_BRIDGE
, RTM_GETLINK
, NULL
, rtnl_bridge_getlink
, NULL
);
3335 rtnl_register(PF_BRIDGE
, RTM_DELLINK
, rtnl_bridge_dellink
, NULL
, NULL
);
3336 rtnl_register(PF_BRIDGE
, RTM_SETLINK
, rtnl_bridge_setlink
, NULL
, NULL
);