projects / project / netifd.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
system-linux: add retry for adding member devices to a bridge
[project/netifd.git] / system-linux.c
1 /*
2 * netifd - network interface daemon
3 * Copyright (C) 2012 Felix Fietkau <nbd@openwrt.org>
4 * Copyright (C) 2013 Jo-Philipp Wich <jow@openwrt.org>
5 * Copyright (C) 2013 Steven Barth <steven@midlink.org>
6 * Copyright (C) 2014 Gioacchino Mazzurco <gio@eigenlab.org>
7 * Copyright (C) 2017 Matthias Schiffer <mschiffer@universe-factory.net>
8 * Copyright (C) 2018 Hans Dedecker <dedeckeh@gmail.com>
9 *
10 * This program is free software; you can redistribute it and/or modify
11 * it under the terms of the GNU General Public License version 2
12 * as published by the Free Software Foundation
13 *
14 * This program is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 * GNU General Public License for more details.
18 */
19 #define _GNU_SOURCE
20
21 #include <sys/socket.h>
22 #include <sys/ioctl.h>
23 #include <sys/stat.h>
24 #include <sys/syscall.h>
25
26 #include <net/if.h>
27 #include <net/if_arp.h>
28
29 #include <limits.h>
30 #include <arpa/inet.h>
31 #include <netinet/ether.h>
32 #include <netinet/in.h>
33
34 #include <linux/rtnetlink.h>
35 #include <linux/neighbour.h>
36 #include <linux/sockios.h>
37 #include <linux/ip.h>
38 #include <linux/if_addr.h>
39 #include <linux/if_link.h>
40 #include <linux/if_vlan.h>
41 #include <linux/if_bridge.h>
42 #include <linux/if_tunnel.h>
43 #include <linux/ip6_tunnel.h>
44 #include <linux/ethtool.h>
45 #include <linux/fib_rules.h>
46 #include <linux/veth.h>
47 #include <linux/version.h>
48
49 #include <sched.h>
50
51 #ifndef RTN_FAILED_POLICY
52 #define RTN_FAILED_POLICY 12
53 #endif
54
55 #ifndef IFA_F_NOPREFIXROUTE
56 #define IFA_F_NOPREFIXROUTE 0x200
57 #endif
58
59 #ifndef IFA_FLAGS
60 #define IFA_FLAGS (IFA_MULTICAST + 1)
61 #endif
62
63 #include <string.h>
64 #include <fcntl.h>
65 #include <glob.h>
66 #include <time.h>
67 #include <unistd.h>
68
69 #include <netlink/msg.h>
70 #include <netlink/attr.h>
71 #include <netlink/socket.h>
72 #include <libubox/uloop.h>
73
74 #include "netifd.h"
75 #include "device.h"
76 #include "system.h"
77 #include "utils.h"
78
79 struct event_socket {
80 struct uloop_fd uloop;
81 struct nl_sock *sock;
82 int bufsize;
83 };
84
85 static int sock_ioctl = -1;
86 static struct nl_sock *sock_rtnl = NULL;
87
88 static int cb_rtnl_event(struct nl_msg *msg, void *arg);
89 static void handle_hotplug_event(struct uloop_fd *u, unsigned int events);
90 static int system_add_proto_tunnel(const char *name, const uint8_t proto,
91 const unsigned int link, struct blob_attr **tb);
92 static int __system_del_ip_tunnel(const char *name, struct blob_attr **tb);
93
94 static char dev_buf[256];
95
96 static void
97 handler_nl_event(struct uloop_fd *u, unsigned int events)
98 {
99 struct event_socket *ev = container_of(u, struct event_socket, uloop);
100 int err;
101 socklen_t errlen = sizeof(err);
102
103 if (!u->error) {
104 nl_recvmsgs_default(ev->sock);
105 return;
106 }
107
108 if (getsockopt(u->fd, SOL_SOCKET, SO_ERROR, (void *)&err, &errlen))
109 goto abort;
110
111 switch(err) {
112 case ENOBUFS:
113 /* Increase rx buffer size on netlink socket */
114 ev->bufsize *= 2;
115 if (nl_socket_set_buffer_size(ev->sock, ev->bufsize, 0))
116 goto abort;
117
118 /* Request full dump since some info got dropped */
119 struct rtgenmsg msg = { .rtgen_family = AF_UNSPEC };
120 nl_send_simple(ev->sock, RTM_GETLINK, NLM_F_DUMP, &msg, sizeof(msg));
121 break;
122
123 default:
124 goto abort;
125 }
126 u->error = false;
127 return;
128
129 abort:
130 uloop_fd_delete(&ev->uloop);
131 return;
132 }
133
134 static struct nl_sock *
135 create_socket(int protocol, int groups)
136 {
137 struct nl_sock *sock;
138
139 sock = nl_socket_alloc();
140 if (!sock)
141 return NULL;
142
143 if (groups)
144 nl_join_groups(sock, groups);
145
146 if (nl_connect(sock, protocol)) {
147 nl_socket_free(sock);
148 return NULL;
149 }
150
151 return sock;
152 }
153
154 static bool
155 create_raw_event_socket(struct event_socket *ev, int protocol, int groups,
156 uloop_fd_handler cb, int flags)
157 {
158 ev->sock = create_socket(protocol, groups);
159 if (!ev->sock)
160 return false;
161
162 ev->uloop.fd = nl_socket_get_fd(ev->sock);
163 ev->uloop.cb = cb;
164 if (uloop_fd_add(&ev->uloop, ULOOP_READ|flags))
165 return false;
166
167 return true;
168 }
169
170 static bool
171 create_event_socket(struct event_socket *ev, int protocol,
172 int (*cb)(struct nl_msg *msg, void *arg))
173 {
174 if (!create_raw_event_socket(ev, protocol, 0, handler_nl_event, ULOOP_ERROR_CB))
175 return false;
176
177 /* Install the valid custom callback handler */
178 nl_socket_modify_cb(ev->sock, NL_CB_VALID, NL_CB_CUSTOM, cb, NULL);
179
180 /* Disable sequence number checking on event sockets */
181 nl_socket_disable_seq_check(ev->sock);
182
183 /* Increase rx buffer size to 65K on event sockets */
184 ev->bufsize = 65535;
185 if (nl_socket_set_buffer_size(ev->sock, ev->bufsize, 0))
186 return false;
187
188 return true;
189 }
190
191 static bool
192 create_hotplug_event_socket(struct event_socket *ev, int protocol,
193 void (*cb)(struct uloop_fd *u, unsigned int events))
194 {
195 if (!create_raw_event_socket(ev, protocol, 1, cb, ULOOP_ERROR_CB))
196 return false;
197
198 /* Increase rx buffer size to 65K on event sockets */
199 ev->bufsize = 65535;
200 if (nl_socket_set_buffer_size(ev->sock, ev->bufsize, 0))
201 return false;
202
203 return true;
204 }
205
206 static bool
207 system_rtn_aton(const char *src, unsigned int *dst)
208 {
209 char *e;
210 unsigned int n;
211
212 if (!strcmp(src, "local"))
213 n = RTN_LOCAL;
214 else if (!strcmp(src, "nat"))
215 n = RTN_NAT;
216 else if (!strcmp(src, "broadcast"))
217 n = RTN_BROADCAST;
218 else if (!strcmp(src, "anycast"))
219 n = RTN_ANYCAST;
220 else if (!strcmp(src, "multicast"))
221 n = RTN_MULTICAST;
222 else if (!strcmp(src, "prohibit"))
223 n = RTN_PROHIBIT;
224 else if (!strcmp(src, "unreachable"))
225 n = RTN_UNREACHABLE;
226 else if (!strcmp(src, "blackhole"))
227 n = RTN_BLACKHOLE;
228 else if (!strcmp(src, "xresolve"))
229 n = RTN_XRESOLVE;
230 else if (!strcmp(src, "unicast"))
231 n = RTN_UNICAST;
232 else if (!strcmp(src, "throw"))
233 n = RTN_THROW;
234 else if (!strcmp(src, "failed_policy"))
235 n = RTN_FAILED_POLICY;
236 else {
237 n = strtoul(src, &e, 0);
238 if (!e || *e || e == src || n > 255)
239 return false;
240 }
241
242 *dst = n;
243 return true;
244 }
245
246 static bool
247 system_tos_aton(const char *src, unsigned *dst)
248 {
249 char *e;
250
251 *dst = strtoul(src, &e, 16);
252 if (e == src || *e || *dst > 255)
253 return false;
254
255 return true;
256 }
257
258 int system_init(void)
259 {
260 static struct event_socket rtnl_event;
261 static struct event_socket hotplug_event;
262
263 sock_ioctl = socket(AF_LOCAL, SOCK_DGRAM, 0);
264 system_fd_set_cloexec(sock_ioctl);
265
266 /* Prepare socket for routing / address control */
267 sock_rtnl = create_socket(NETLINK_ROUTE, 0);
268 if (!sock_rtnl)
269 return -1;
270
271 if (!create_event_socket(&rtnl_event, NETLINK_ROUTE, cb_rtnl_event))
272 return -1;
273
274 if (!create_hotplug_event_socket(&hotplug_event, NETLINK_KOBJECT_UEVENT,
275 handle_hotplug_event))
276 return -1;
277
278 /* Receive network link events form kernel */
279 nl_socket_add_membership(rtnl_event.sock, RTNLGRP_LINK);
280
281 return 0;
282 }
283
284 static void system_set_sysctl(const char *path, const char *val)
285 {
286 int fd;
287
288 fd = open(path, O_WRONLY);
289 if (fd < 0)
290 return;
291
292 if (write(fd, val, strlen(val))) {}
293 close(fd);
294 }
295
296 static void system_set_dev_sysctl(const char *path, const char *device, const char *val)
297 {
298 snprintf(dev_buf, sizeof(dev_buf), path, device);
299 system_set_sysctl(dev_buf, val);
300 }
301
302 static void system_set_disable_ipv6(struct device *dev, const char *val)
303 {
304 system_set_dev_sysctl("/proc/sys/net/ipv6/conf/%s/disable_ipv6", dev->ifname, val);
305 }
306
307 static void system_set_rpfilter(struct device *dev, const char *val)
308 {
309 system_set_dev_sysctl("/proc/sys/net/ipv4/conf/%s/rp_filter", dev->ifname, val);
310 }
311
312 static void system_set_acceptlocal(struct device *dev, const char *val)
313 {
314 system_set_dev_sysctl("/proc/sys/net/ipv4/conf/%s/accept_local", dev->ifname, val);
315 }
316
317 static void system_set_igmpversion(struct device *dev, const char *val)
318 {
319 system_set_dev_sysctl("/proc/sys/net/ipv4/conf/%s/force_igmp_version", dev->ifname, val);
320 }
321
322 static void system_set_mldversion(struct device *dev, const char *val)
323 {
324 system_set_dev_sysctl("/proc/sys/net/ipv6/conf/%s/force_mld_version", dev->ifname, val);
325 }
326
327 static void system_set_neigh4reachabletime(struct device *dev, const char *val)
328 {
329 system_set_dev_sysctl("/proc/sys/net/ipv4/neigh/%s/base_reachable_time_ms", dev->ifname, val);
330 }
331
332 static void system_set_neigh6reachabletime(struct device *dev, const char *val)
333 {
334 system_set_dev_sysctl("/proc/sys/net/ipv6/neigh/%s/base_reachable_time_ms", dev->ifname, val);
335 }
336
337 static void system_set_neigh4gcstaletime(struct device *dev, const char *val)
338 {
339 system_set_dev_sysctl("/proc/sys/net/ipv4/neigh/%s/gc_stale_time", dev->ifname, val);
340 }
341
342 static void system_set_neigh6gcstaletime(struct device *dev, const char *val)
343 {
344 system_set_dev_sysctl("/proc/sys/net/ipv6/neigh/%s/gc_stale_time", dev->ifname, val);
345 }
346
347 static void system_set_neigh4locktime(struct device *dev, const char *val)
348 {
349 system_set_dev_sysctl("/proc/sys/net/ipv4/neigh/%s/locktime", dev->ifname, val);
350 }
351
352 static void system_set_dadtransmits(struct device *dev, const char *val)
353 {
354 system_set_dev_sysctl("/proc/sys/net/ipv6/conf/%s/dad_transmits", dev->ifname, val);
355 }
356
357 static void system_bridge_set_multicast_to_unicast(struct device *dev, const char *val)
358 {
359 system_set_dev_sysctl("/sys/class/net/%s/brport/multicast_to_unicast", dev->ifname, val);
360 }
361
362 static void system_bridge_set_multicast_fast_leave(struct device *dev, const char *val)
363 {
364 system_set_dev_sysctl("/sys/class/net/%s/brport/multicast_fast_leave", dev->ifname, val);
365 }
366
367 static void system_bridge_set_hairpin_mode(struct device *dev, const char *val)
368 {
369 system_set_dev_sysctl("/sys/class/net/%s/brport/hairpin_mode", dev->ifname, val);
370 }
371
372 static void system_bridge_set_isolated(struct device *dev, const char *val)
373 {
374 system_set_dev_sysctl("/sys/class/net/%s/brport/isolated", dev->ifname, val);
375 }
376
377 static void system_bridge_set_multicast_router(struct device *dev, const char *val, bool bridge)
378 {
379 system_set_dev_sysctl(bridge ? "/sys/class/net/%s/bridge/multicast_router" :
380 "/sys/class/net/%s/brport/multicast_router",
381 dev->ifname, val);
382 }
383
384 static void system_bridge_set_robustness(struct device *dev, const char *val)
385 {
386 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_startup_query_count",
387 dev->ifname, val);
388 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_last_member_count",
389 dev->ifname, val);
390 }
391
392 static void system_bridge_set_query_interval(struct device *dev, const char *val)
393 {
394 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_query_interval",
395 dev->ifname, val);
396 }
397
398 static void system_bridge_set_query_response_interval(struct device *dev, const char *val)
399 {
400 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_query_response_interval",
401 dev->ifname, val);
402 }
403
404 static void system_bridge_set_last_member_interval(struct device *dev, const char *val)
405 {
406 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_last_member_interval",
407 dev->ifname, val);
408 }
409
410 static void system_bridge_set_membership_interval(struct device *dev, const char *val)
411 {
412 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_membership_interval",
413 dev->ifname, val);
414 }
415
416 static void system_bridge_set_other_querier_timeout(struct device *dev, const char *val)
417 {
418 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_querier_interval",
419 dev->ifname, val);
420 }
421
422 static void system_bridge_set_startup_query_interval(struct device *dev, const char *val)
423 {
424 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_startup_query_interval",
425 dev->ifname, val);
426 }
427
428 static void system_bridge_set_stp_state(struct device *dev, const char *val)
429 {
430 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/stp_state", dev->ifname, val);
431 }
432
433 static void system_bridge_set_forward_delay(struct device *dev, const char *val)
434 {
435 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/forward_delay", dev->ifname, val);
436 }
437
438 static void system_bridge_set_priority(struct device *dev, const char *val)
439 {
440 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/priority", dev->ifname, val);
441 }
442
443 static void system_bridge_set_ageing_time(struct device *dev, const char *val)
444 {
445 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/ageing_time", dev->ifname, val);
446 }
447
448 static void system_bridge_set_hello_time(struct device *dev, const char *val)
449 {
450 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/hello_time", dev->ifname, val);
451 }
452
453 static void system_bridge_set_max_age(struct device *dev, const char *val)
454 {
455 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/max_age", dev->ifname, val);
456 }
457
458 static void system_bridge_set_learning(struct device *dev, const char *val)
459 {
460 system_set_dev_sysctl("/sys/class/net/%s/brport/learning", dev->ifname, val);
461 }
462
463 static void system_bridge_set_unicast_flood(struct device *dev, const char *val)
464 {
465 system_set_dev_sysctl("/sys/class/net/%s/brport/unicast_flood", dev->ifname, val);
466 }
467
468 static void system_set_sendredirects(struct device *dev, const char *val)
469 {
470 system_set_dev_sysctl("/proc/sys/net/ipv4/conf/%s/send_redirects", dev->ifname, val);
471 }
472
473 static void system_bridge_set_vlan_filtering(struct device *dev, const char *val)
474 {
475 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/vlan_filtering", dev->ifname, val);
476 }
477
478 static int system_get_sysctl(const char *path, char *buf, const size_t buf_sz)
479 {
480 int fd = -1, ret = -1;
481
482 fd = open(path, O_RDONLY);
483 if (fd < 0)
484 goto out;
485
486 ssize_t len = read(fd, buf, buf_sz - 1);
487 if (len < 0)
488 goto out;
489
490 ret = buf[len] = 0;
491
492 out:
493 if (fd >= 0)
494 close(fd);
495
496 return ret;
497 }
498
499 static int
500 system_get_dev_sysctl(const char *path, const char *device, char *buf, const size_t buf_sz)
501 {
502 snprintf(dev_buf, sizeof(dev_buf), path, device);
503 return system_get_sysctl(dev_buf, buf, buf_sz);
504 }
505
506 static int system_get_disable_ipv6(struct device *dev, char *buf, const size_t buf_sz)
507 {
508 return system_get_dev_sysctl("/proc/sys/net/ipv6/conf/%s/disable_ipv6",
509 dev->ifname, buf, buf_sz);
510 }
511
512 static int system_get_rpfilter(struct device *dev, char *buf, const size_t buf_sz)
513 {
514 return system_get_dev_sysctl("/proc/sys/net/ipv4/conf/%s/rp_filter",
515 dev->ifname, buf, buf_sz);
516 }
517
518 static int system_get_acceptlocal(struct device *dev, char *buf, const size_t buf_sz)
519 {
520 return system_get_dev_sysctl("/proc/sys/net/ipv4/conf/%s/accept_local",
521 dev->ifname, buf, buf_sz);
522 }
523
524 static int system_get_igmpversion(struct device *dev, char *buf, const size_t buf_sz)
525 {
526 return system_get_dev_sysctl("/proc/sys/net/ipv4/conf/%s/force_igmp_version",
527 dev->ifname, buf, buf_sz);
528 }
529
530 static int system_get_mldversion(struct device *dev, char *buf, const size_t buf_sz)
531 {
532 return system_get_dev_sysctl("/proc/sys/net/ipv6/conf/%s/force_mld_version",
533 dev->ifname, buf, buf_sz);
534 }
535
536 static int system_get_neigh4reachabletime(struct device *dev, char *buf, const size_t buf_sz)
537 {
538 return system_get_dev_sysctl("/proc/sys/net/ipv4/neigh/%s/base_reachable_time_ms",
539 dev->ifname, buf, buf_sz);
540 }
541
542 static int system_get_neigh6reachabletime(struct device *dev, char *buf, const size_t buf_sz)
543 {
544 return system_get_dev_sysctl("/proc/sys/net/ipv6/neigh/%s/base_reachable_time_ms",
545 dev->ifname, buf, buf_sz);
546 }
547
548 static int system_get_neigh4gcstaletime(struct device *dev, char *buf, const size_t buf_sz)
549 {
550 return system_get_dev_sysctl("/proc/sys/net/ipv4/neigh/%s/gc_stale_time",
551 dev->ifname, buf, buf_sz);
552 }
553
554 static int system_get_neigh6gcstaletime(struct device *dev, char *buf, const size_t buf_sz)
555 {
556 return system_get_dev_sysctl("/proc/sys/net/ipv6/neigh/%s/gc_stale_time",
557 dev->ifname, buf, buf_sz);
558 }
559
560 static int system_get_neigh4locktime(struct device *dev, char *buf, const size_t buf_sz)
561 {
562 return system_get_dev_sysctl("/proc/sys/net/ipv4/neigh/%s/locktime",
563 dev->ifname, buf, buf_sz);
564 }
565
566 static int system_get_dadtransmits(struct device *dev, char *buf, const size_t buf_sz)
567 {
568 return system_get_dev_sysctl("/proc/sys/net/ipv6/conf/%s/dad_transmits",
569 dev->ifname, buf, buf_sz);
570 }
571
572 static int system_get_sendredirects(struct device *dev, char *buf, const size_t buf_sz)
573 {
574 return system_get_dev_sysctl("/proc/sys/net/ipv4/conf/%s/send_redirects",
575 dev->ifname, buf, buf_sz);
576 }
577
578 /* Evaluate netlink messages */
579 static int cb_rtnl_event(struct nl_msg *msg, void *arg)
580 {
581 struct nlmsghdr *nh = nlmsg_hdr(msg);
582 struct nlattr *nla[__IFLA_MAX];
583 int link_state = 0;
584 char buf[10];
585
586 if (nh->nlmsg_type != RTM_NEWLINK)
587 goto out;
588
589 nlmsg_parse(nh, sizeof(struct ifinfomsg), nla, __IFLA_MAX - 1, NULL);
590 if (!nla[IFLA_IFNAME])
591 goto out;
592
593 struct device *dev = device_find(nla_data(nla[IFLA_IFNAME]));
594 if (!dev)
595 goto out;
596
597 if (!system_get_dev_sysctl("/sys/class/net/%s/carrier", dev->ifname, buf, sizeof(buf)))
598 link_state = strtoul(buf, NULL, 0);
599
600 if (dev->type == &simple_device_type)
601 device_set_present(dev, true);
602
603 device_set_link(dev, link_state ? true : false);
604
605 out:
606 return 0;
607 }
608
609 static void
610 handle_hotplug_msg(char *data, int size)
611 {
612 const char *subsystem = NULL, *interface = NULL, *interface_old = NULL;
613 char *cur, *end, *sep;
614 struct device *dev;
615 int skip;
616 bool add, move = false;
617
618 if (!strncmp(data, "add@", 4))
619 add = true;
620 else if (!strncmp(data, "remove@", 7))
621 add = false;
622 else if (!strncmp(data, "move@", 5)) {
623 add = true;
624 move = true;
625 }
626 else
627 return;
628
629 skip = strlen(data) + 1;
630 end = data + size;
631
632 for (cur = data + skip; cur < end; cur += skip) {
633 skip = strlen(cur) + 1;
634
635 sep = strchr(cur, '=');
636 if (!sep)
637 continue;
638
639 *sep = 0;
640 if (!strcmp(cur, "INTERFACE"))
641 interface = sep + 1;
642 else if (!strcmp(cur, "SUBSYSTEM")) {
643 subsystem = sep + 1;
644 if (strcmp(subsystem, "net") != 0)
645 return;
646 } else if (!strcmp(cur, "DEVPATH_OLD")) {
647 interface_old = strrchr(sep + 1, '/');
648 if (interface_old)
649 interface_old++;
650 }
651 }
652
653 if (subsystem && interface) {
654 if (move && interface_old)
655 goto move;
656 else
657 goto found;
658 }
659
660 return;
661
662 move:
663 dev = device_find(interface_old);
664 if (!dev)
665 return;
666
667 if (dev->type != &simple_device_type)
668 goto found;
669
670 device_set_present(dev, false);
671
672 return;
673
674 found:
675 dev = device_find(interface);
676 if (!dev)
677 return;
678
679 if (dev->type != &simple_device_type)
680 return;
681
682 device_set_present(dev, add);
683 }
684
685 static void
686 handle_hotplug_event(struct uloop_fd *u, unsigned int events)
687 {
688 struct event_socket *ev = container_of(u, struct event_socket, uloop);
689 struct sockaddr_nl nla;
690 unsigned char *buf = NULL;
691 int size;
692 int err;
693 socklen_t errlen = sizeof(err);
694
695 if (!u->error) {
696 while ((size = nl_recv(ev->sock, &nla, &buf, NULL)) > 0) {
697 if (nla.nl_pid == 0)
698 handle_hotplug_msg((char *) buf, size);
699
700 free(buf);
701 }
702 return;
703 }
704
705 if (getsockopt(u->fd, SOL_SOCKET, SO_ERROR, (void *)&err, &errlen))
706 goto abort;
707
708 switch(err) {
709 case ENOBUFS:
710 /* Increase rx buffer size on netlink socket */
711 ev->bufsize *= 2;
712 if (nl_socket_set_buffer_size(ev->sock, ev->bufsize, 0))
713 goto abort;
714 break;
715
716 default:
717 goto abort;
718 }
719 u->error = false;
720 return;
721
722 abort:
723 uloop_fd_delete(&ev->uloop);
724 return;
725 }
726
727 static int system_rtnl_call(struct nl_msg *msg)
728 {
729 int ret;
730
731 ret = nl_send_auto_complete(sock_rtnl, msg);
732 nlmsg_free(msg);
733
734 if (ret < 0)
735 return ret;
736
737 return nl_wait_for_ack(sock_rtnl);
738 }
739
740 int system_bridge_delbr(struct device *bridge)
741 {
742 return ioctl(sock_ioctl, SIOCBRDELBR, bridge->ifname);
743 }
744
745 static int system_bridge_if(const char *bridge, struct device *dev, int cmd, void *data)
746 {
747 struct ifreq ifr;
748
749 memset(&ifr, 0, sizeof(ifr));
750 if (dev)
751 ifr.ifr_ifindex = dev->ifindex;
752 else
753 ifr.ifr_data = data;
754 strncpy(ifr.ifr_name, bridge, sizeof(ifr.ifr_name) - 1);
755 return ioctl(sock_ioctl, cmd, &ifr);
756 }
757
758 static bool system_is_bridge(const char *name, char *buf, int buflen)
759 {
760 struct stat st;
761
762 snprintf(buf, buflen, "/sys/devices/virtual/net/%s/bridge", name);
763 if (stat(buf, &st) < 0)
764 return false;
765
766 return true;
767 }
768
769 static char *system_get_bridge(const char *name, char *buf, int buflen)
770 {
771 char *path;
772 ssize_t len = -1;
773 glob_t gl;
774
775 snprintf(buf, buflen, "/sys/devices/virtual/net/*/brif/%s/bridge", name);
776 if (glob(buf, GLOB_NOSORT, NULL, &gl) < 0)
777 return NULL;
778
779 if (gl.gl_pathc > 0)
780 len = readlink(gl.gl_pathv[0], buf, buflen);
781
782 globfree(&gl);
783
784 if (len < 0)
785 return NULL;
786
787 buf[len] = 0;
788 path = strrchr(buf, '/');
789 if (!path)
790 return NULL;
791
792 return path + 1;
793 }
794
795 static void
796 system_bridge_set_wireless(struct device *bridge, struct device *dev)
797 {
798 bool mcast_to_ucast = dev->wireless_ap;
799 bool hairpin = true;
800
801 if (bridge->settings.flags & DEV_OPT_MULTICAST_TO_UNICAST &&
802 !bridge->settings.multicast_to_unicast)
803 mcast_to_ucast = false;
804
805 if (!mcast_to_ucast || dev->wireless_isolate)
806 hairpin = false;
807
808 system_bridge_set_multicast_to_unicast(dev, mcast_to_ucast ? "1" : "0");
809 system_bridge_set_hairpin_mode(dev, hairpin ? "1" : "0");
810 }
811
812 int system_bridge_addif(struct device *bridge, struct device *dev)
813 {
814 char buf[64];
815 char *oldbr;
816 int tries = 0;
817 int ret;
818
819 retry:
820 ret = 0;
821 oldbr = system_get_bridge(dev->ifname, dev_buf, sizeof(dev_buf));
822 if (!oldbr || strcmp(oldbr, bridge->ifname) != 0) {
823 ret = system_bridge_if(bridge->ifname, dev, SIOCBRADDIF, NULL);
824 tries++;
825 D(SYSTEM, "Failed to add device '%s' to bridge '%s' (tries=%d): %s\n",
826 dev->ifname, bridge->ifname, tries, strerror(errno));
827 if (tries <= 3)
828 goto retry;
829 }
830
831 if (dev->wireless)
832 system_bridge_set_wireless(bridge, dev);
833
834 if (dev->settings.flags & DEV_OPT_MULTICAST_ROUTER) {
835 snprintf(buf, sizeof(buf), "%u", dev->settings.multicast_router);
836 system_bridge_set_multicast_router(dev, buf, false);
837 }
838
839 if (dev->settings.flags & DEV_OPT_MULTICAST_FAST_LEAVE &&
840 dev->settings.multicast_fast_leave)
841 system_bridge_set_multicast_fast_leave(dev, "1");
842
843 if (dev->settings.flags & DEV_OPT_LEARNING &&
844 !dev->settings.learning)
845 system_bridge_set_learning(dev, "0");
846
847 if (dev->settings.flags & DEV_OPT_UNICAST_FLOOD &&
848 !dev->settings.unicast_flood)
849 system_bridge_set_unicast_flood(dev, "0");
850
851 if (dev->settings.flags & DEV_OPT_ISOLATE &&
852 dev->settings.isolate)
853 system_bridge_set_isolated(dev, "1");
854
855 return ret;
856 }
857
858 int system_bridge_delif(struct device *bridge, struct device *dev)
859 {
860 return system_bridge_if(bridge->ifname, dev, SIOCBRDELIF, NULL);
861 }
862
863 int system_bridge_vlan(const char *iface, uint16_t vid, bool add, unsigned int vflags)
864 {
865 struct ifinfomsg ifi = { .ifi_family = PF_BRIDGE, };
866 struct bridge_vlan_info vinfo = { .vid = vid, };
867 unsigned short flags = 0;
868 struct nlattr *afspec;
869 struct nl_msg *nlm;
870 int ret = 0;
871
872 ifi.ifi_index = if_nametoindex(iface);
873 if (!ifi.ifi_index)
874 return -1;
875
876 nlm = nlmsg_alloc_simple(add ? RTM_SETLINK : RTM_DELLINK, NLM_F_REQUEST);
877 if (!nlm)
878 return -1;
879
880 nlmsg_append(nlm, &ifi, sizeof(ifi), 0);
881
882 if (vflags & BRVLAN_F_SELF)
883 flags |= BRIDGE_FLAGS_SELF;
884
885 if (vflags & BRVLAN_F_PVID)
886 vinfo.flags |= BRIDGE_VLAN_INFO_PVID;
887
888 if (vflags & BRVLAN_F_UNTAGGED)
889 vinfo.flags |= BRIDGE_VLAN_INFO_UNTAGGED;
890
891 afspec = nla_nest_start(nlm, IFLA_AF_SPEC);
892 if (!afspec) {
893 ret = -ENOMEM;
894 goto failure;
895 }
896
897 if (flags)
898 nla_put_u16(nlm, IFLA_BRIDGE_FLAGS, flags);
899
900 nla_put(nlm, IFLA_BRIDGE_VLAN_INFO, sizeof(vinfo), &vinfo);
901 nla_nest_end(nlm, afspec);
902
903 return system_rtnl_call(nlm);
904
905 failure:
906 nlmsg_free(nlm);
907 return ret;
908 }
909
910 int system_if_resolve(struct device *dev)
911 {
912 struct ifreq ifr;
913
914 memset(&ifr, 0, sizeof(ifr));
915 strncpy(ifr.ifr_name, dev->ifname, sizeof(ifr.ifr_name) - 1);
916 if (!ioctl(sock_ioctl, SIOCGIFINDEX, &ifr))
917 return ifr.ifr_ifindex;
918 else
919 return 0;
920 }
921
922 static int system_if_flags(const char *ifname, unsigned add, unsigned rem)
923 {
924 struct ifreq ifr;
925
926 memset(&ifr, 0, sizeof(ifr));
927 strncpy(ifr.ifr_name, ifname, sizeof(ifr.ifr_name) - 1);
928 if (ioctl(sock_ioctl, SIOCGIFFLAGS, &ifr) < 0)
929 return -1;
930
931 ifr.ifr_flags |= add;
932 ifr.ifr_flags &= ~rem;
933 return ioctl(sock_ioctl, SIOCSIFFLAGS, &ifr);
934 }
935
936 struct clear_data {
937 struct nl_msg *msg;
938 struct device *dev;
939 int type;
940 int size;
941 int af;
942 };
943
944
945 static bool check_ifaddr(struct nlmsghdr *hdr, int ifindex)
946 {
947 struct ifaddrmsg *ifa = NLMSG_DATA(hdr);
948
949 return ifa->ifa_index == ifindex;
950 }
951
952 static bool check_route(struct nlmsghdr *hdr, int ifindex)
953 {
954 struct rtmsg *r = NLMSG_DATA(hdr);
955 struct nlattr *tb[__RTA_MAX];
956
957 if (r->rtm_protocol == RTPROT_KERNEL &&
958 r->rtm_family == AF_INET6)
959 return false;
960
961 nlmsg_parse(hdr, sizeof(struct rtmsg), tb, __RTA_MAX - 1, NULL);
962 if (!tb[RTA_OIF])
963 return false;
964
965 return *(int *)RTA_DATA(tb[RTA_OIF]) == ifindex;
966 }
967
968 static bool check_rule(struct nlmsghdr *hdr, int ifindex)
969 {
970 return true;
971 }
972
973 static int cb_clear_event(struct nl_msg *msg, void *arg)
974 {
975 struct clear_data *clr = arg;
976 struct nlmsghdr *hdr = nlmsg_hdr(msg);
977 bool (*cb)(struct nlmsghdr *, int ifindex);
978 int type, ret;
979
980 switch(clr->type) {
981 case RTM_GETADDR:
982 type = RTM_DELADDR;
983 if (hdr->nlmsg_type != RTM_NEWADDR)
984 return NL_SKIP;
985
986 cb = check_ifaddr;
987 break;
988 case RTM_GETROUTE:
989 type = RTM_DELROUTE;
990 if (hdr->nlmsg_type != RTM_NEWROUTE)
991 return NL_SKIP;
992
993 cb = check_route;
994 break;
995 case RTM_GETRULE:
996 type = RTM_DELRULE;
997 if (hdr->nlmsg_type != RTM_NEWRULE)
998 return NL_SKIP;
999
1000 cb = check_rule;
1001 break;
1002 default:
1003 return NL_SKIP;
1004 }
1005
1006 if (!cb(hdr, clr->dev ? clr->dev->ifindex : 0))
1007 return NL_SKIP;
1008
1009 if (type == RTM_DELRULE)
1010 D(SYSTEM, "Remove a rule\n");
1011 else
1012 D(SYSTEM, "Remove %s from device %s\n",
1013 type == RTM_DELADDR ? "an address" : "a route",
1014 clr->dev->ifname);
1015
1016 memcpy(nlmsg_hdr(clr->msg), hdr, hdr->nlmsg_len);
1017 hdr = nlmsg_hdr(clr->msg);
1018 hdr->nlmsg_type = type;
1019 hdr->nlmsg_flags = NLM_F_REQUEST;
1020
1021 nl_socket_disable_auto_ack(sock_rtnl);
1022 ret = nl_send_auto_complete(sock_rtnl, clr->msg);
1023 if (ret < 0) {
1024 if (type == RTM_DELRULE)
1025 D(SYSTEM, "Error deleting a rule: %d\n", ret);
1026 else
1027 D(SYSTEM, "Error deleting %s from device '%s': %d\n",
1028 type == RTM_DELADDR ? "an address" : "a route",
1029 clr->dev->ifname, ret);
1030 }
1031
1032 nl_socket_enable_auto_ack(sock_rtnl);
1033
1034 return NL_SKIP;
1035 }
1036
1037 static int
1038 cb_finish_event(struct nl_msg *msg, void *arg)
1039 {
1040 int *pending = arg;
1041 *pending = 0;
1042 return NL_STOP;
1043 }
1044
1045 static int
1046 error_handler(struct sockaddr_nl *nla, struct nlmsgerr *err, void *arg)
1047 {
1048 int *pending = arg;
1049 *pending = err->error;
1050 return NL_STOP;
1051 }
1052
1053 static void
1054 system_if_clear_entries(struct device *dev, int type, int af)
1055 {
1056 struct clear_data clr;
1057 struct nl_cb *cb;
1058 struct rtmsg rtm = {
1059 .rtm_family = af,
1060 .rtm_flags = RTM_F_CLONED,
1061 };
1062 int flags = NLM_F_DUMP;
1063 int pending = 1;
1064
1065 clr.af = af;
1066 clr.dev = dev;
1067 clr.type = type;
1068 switch (type) {
1069 case RTM_GETADDR:
1070 case RTM_GETRULE:
1071 clr.size = sizeof(struct rtgenmsg);
1072 break;
1073 case RTM_GETROUTE:
1074 clr.size = sizeof(struct rtmsg);
1075 break;
1076 default:
1077 return;
1078 }
1079
1080 cb = nl_cb_alloc(NL_CB_DEFAULT);
1081 if (!cb)
1082 return;
1083
1084 clr.msg = nlmsg_alloc_simple(type, flags);
1085 if (!clr.msg)
1086 goto out;
1087
1088 nlmsg_append(clr.msg, &rtm, clr.size, 0);
1089 nl_cb_set(cb, NL_CB_VALID, NL_CB_CUSTOM, cb_clear_event, &clr);
1090 nl_cb_set(cb, NL_CB_FINISH, NL_CB_CUSTOM, cb_finish_event, &pending);
1091 nl_cb_err(cb, NL_CB_CUSTOM, error_handler, &pending);
1092
1093 if (nl_send_auto_complete(sock_rtnl, clr.msg) < 0)
1094 goto free;
1095
1096 while (pending > 0)
1097 nl_recvmsgs(sock_rtnl, cb);
1098
1099 free:
1100 nlmsg_free(clr.msg);
1101 out:
1102 nl_cb_put(cb);
1103 }
1104
1105 /*
1106 * Clear bridge (membership) state and bring down device
1107 */
1108 void system_if_clear_state(struct device *dev)
1109 {
1110 static char buf[256];
1111 char *bridge;
1112 device_set_ifindex(dev, system_if_resolve(dev));
1113
1114 if (dev->external || !dev->ifindex)
1115 return;
1116
1117 system_if_flags(dev->ifname, 0, IFF_UP);
1118
1119 if (system_is_bridge(dev->ifname, buf, sizeof(buf))) {
1120 D(SYSTEM, "Delete existing bridge named '%s'\n", dev->ifname);
1121 system_bridge_delbr(dev);
1122 return;
1123 }
1124
1125 bridge = system_get_bridge(dev->ifname, buf, sizeof(buf));
1126 if (bridge) {
1127 D(SYSTEM, "Remove device '%s' from bridge '%s'\n", dev->ifname, bridge);
1128 system_bridge_if(bridge, dev, SIOCBRDELIF, NULL);
1129 }
1130
1131 system_if_clear_entries(dev, RTM_GETROUTE, AF_INET);
1132 system_if_clear_entries(dev, RTM_GETADDR, AF_INET);
1133 system_if_clear_entries(dev, RTM_GETROUTE, AF_INET6);
1134 system_if_clear_entries(dev, RTM_GETADDR, AF_INET6);
1135 system_if_clear_entries(dev, RTM_GETNEIGH, AF_INET);
1136 system_if_clear_entries(dev, RTM_GETNEIGH, AF_INET6);
1137 system_set_disable_ipv6(dev, "0");
1138 }
1139
1140 static inline unsigned long
1141 sec_to_jiffies(int val)
1142 {
1143 return (unsigned long) val * 100;
1144 }
1145
1146 static void system_bridge_conf_multicast_deps(struct device *bridge,
1147 struct bridge_config *cfg,
1148 char *buf,
1149 int buf_len)
1150 {
1151 int val;
1152
1153 if (cfg->flags & BRIDGE_OPT_ROBUSTNESS ||
1154 cfg->flags & BRIDGE_OPT_QUERY_INTERVAL ||
1155 cfg->flags & BRIDGE_OPT_QUERY_RESPONSE_INTERVAL) {
1156 val = cfg->robustness * cfg->query_interval +
1157 cfg->query_response_interval;
1158
1159 snprintf(buf, buf_len, "%i", val);
1160 system_bridge_set_membership_interval(bridge, buf);
1161
1162 val = cfg->robustness * cfg->query_interval +
1163 cfg->query_response_interval / 2;
1164
1165 snprintf(buf, buf_len, "%i", val);
1166 system_bridge_set_other_querier_timeout(bridge, buf);
1167 }
1168
1169 if (cfg->flags & BRIDGE_OPT_QUERY_INTERVAL) {
1170 val = cfg->query_interval / 4;
1171
1172 snprintf(buf, buf_len, "%i", val);
1173 system_bridge_set_startup_query_interval(bridge, buf);
1174 }
1175 }
1176
1177 static void system_bridge_conf_multicast(struct device *bridge,
1178 struct bridge_config *cfg,
1179 char *buf,
1180 int buf_len)
1181 {
1182 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_snooping",
1183 bridge->ifname, cfg->igmp_snoop ? "1" : "0");
1184
1185 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/multicast_querier",
1186 bridge->ifname, cfg->multicast_querier ? "1" : "0");
1187
1188 snprintf(buf, buf_len, "%i", cfg->hash_max);
1189 system_set_dev_sysctl("/sys/devices/virtual/net/%s/bridge/hash_max",
1190 bridge->ifname, buf);
1191
1192 if (bridge->settings.flags & DEV_OPT_MULTICAST_ROUTER) {
1193 snprintf(buf, buf_len, "%u", bridge->settings.multicast_router);
1194 system_bridge_set_multicast_router(bridge, buf, true);
1195 }
1196
1197 if (cfg->flags & BRIDGE_OPT_ROBUSTNESS) {
1198 snprintf(buf, buf_len, "%i", cfg->robustness);
1199 system_bridge_set_robustness(bridge, buf);
1200 }
1201
1202 if (cfg->flags & BRIDGE_OPT_QUERY_INTERVAL) {
1203 snprintf(buf, buf_len, "%i", cfg->query_interval);
1204 system_bridge_set_query_interval(bridge, buf);
1205 }
1206
1207 if (cfg->flags & BRIDGE_OPT_QUERY_RESPONSE_INTERVAL) {
1208 snprintf(buf, buf_len, "%i", cfg->query_response_interval);
1209 system_bridge_set_query_response_interval(bridge, buf);
1210 }
1211
1212 if (cfg->flags & BRIDGE_OPT_LAST_MEMBER_INTERVAL) {
1213 snprintf(buf, buf_len, "%i", cfg->last_member_interval);
1214 system_bridge_set_last_member_interval(bridge, buf);
1215 }
1216
1217 system_bridge_conf_multicast_deps(bridge, cfg, buf, buf_len);
1218 }
1219
1220 int system_bridge_addbr(struct device *bridge, struct bridge_config *cfg)
1221 {
1222 char buf[64];
1223
1224 if (ioctl(sock_ioctl, SIOCBRADDBR, bridge->ifname) < 0)
1225 return -1;
1226
1227 system_bridge_set_stp_state(bridge, cfg->stp ? "1" : "0");
1228
1229 snprintf(buf, sizeof(buf), "%lu", sec_to_jiffies(cfg->forward_delay));
1230 system_bridge_set_forward_delay(bridge, buf);
1231
1232 system_bridge_conf_multicast(bridge, cfg, buf, sizeof(buf));
1233 system_bridge_set_vlan_filtering(bridge, cfg->vlan_filtering ? "1" : "0");
1234
1235 snprintf(buf, sizeof(buf), "%d", cfg->priority);
1236 system_bridge_set_priority(bridge, buf);
1237
1238 if (cfg->flags & BRIDGE_OPT_AGEING_TIME) {
1239 snprintf(buf, sizeof(buf), "%lu", sec_to_jiffies(cfg->ageing_time));
1240 system_bridge_set_ageing_time(bridge, buf);
1241 }
1242
1243 if (cfg->flags & BRIDGE_OPT_HELLO_TIME) {
1244 snprintf(buf, sizeof(buf), "%lu", sec_to_jiffies(cfg->hello_time));
1245 system_bridge_set_hello_time(bridge, buf);
1246 }
1247
1248 if (cfg->flags & BRIDGE_OPT_MAX_AGE) {
1249 snprintf(buf, sizeof(buf), "%lu", sec_to_jiffies(cfg->max_age));
1250 system_bridge_set_max_age(bridge, buf);
1251 }
1252
1253 return 0;
1254 }
1255
1256 int system_macvlan_add(struct device *macvlan, struct device *dev, struct macvlan_config *cfg)
1257 {
1258 struct nl_msg *msg;
1259 struct nlattr *linkinfo, *data;
1260 struct ifinfomsg iim = { .ifi_family = AF_UNSPEC, };
1261 int i, rv;
1262 static const struct {
1263 const char *name;
1264 enum macvlan_mode val;
1265 } modes[] = {
1266 { "private", MACVLAN_MODE_PRIVATE },
1267 { "vepa", MACVLAN_MODE_VEPA },
1268 { "bridge", MACVLAN_MODE_BRIDGE },
1269 { "passthru", MACVLAN_MODE_PASSTHRU },
1270 };
1271
1272 msg = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_CREATE | NLM_F_EXCL);
1273
1274 if (!msg)
1275 return -1;
1276
1277 nlmsg_append(msg, &iim, sizeof(iim), 0);
1278
1279 if (cfg->flags & MACVLAN_OPT_MACADDR)
1280 nla_put(msg, IFLA_ADDRESS, sizeof(cfg->macaddr), cfg->macaddr);
1281 nla_put_string(msg, IFLA_IFNAME, macvlan->ifname);
1282 nla_put_u32(msg, IFLA_LINK, dev->ifindex);
1283
1284 if (!(linkinfo = nla_nest_start(msg, IFLA_LINKINFO)))
1285 goto nla_put_failure;
1286
1287 nla_put_string(msg, IFLA_INFO_KIND, "macvlan");
1288
1289 if (!(data = nla_nest_start(msg, IFLA_INFO_DATA)))
1290 goto nla_put_failure;
1291
1292 if (cfg->mode) {
1293 for (i = 0; i < ARRAY_SIZE(modes); i++) {
1294 if (strcmp(cfg->mode, modes[i].name) != 0)
1295 continue;
1296
1297 nla_put_u32(msg, IFLA_MACVLAN_MODE, modes[i].val);
1298 break;
1299 }
1300 }
1301
1302 nla_nest_end(msg, data);
1303 nla_nest_end(msg, linkinfo);
1304
1305 rv = system_rtnl_call(msg);
1306 if (rv)
1307 D(SYSTEM, "Error adding macvlan '%s' over '%s': %d\n", macvlan->ifname, dev->ifname, rv);
1308
1309 return rv;
1310
1311 nla_put_failure:
1312 nlmsg_free(msg);
1313 return -ENOMEM;
1314 }
1315
1316 int system_link_netns_move(struct device *dev, int netns_fd, const char *target_ifname)
1317 {
1318 struct nl_msg *msg;
1319 struct ifinfomsg iim = {
1320 .ifi_family = AF_UNSPEC,
1321 };
1322
1323 if (!dev)
1324 return -1;
1325
1326 iim.ifi_index = system_if_resolve(dev);
1327 msg = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST);
1328
1329 if (!msg)
1330 return -1;
1331
1332 nlmsg_append(msg, &iim, sizeof(iim), 0);
1333 if (target_ifname)
1334 nla_put_string(msg, IFLA_IFNAME, target_ifname);
1335
1336 nla_put_u32(msg, IFLA_NET_NS_FD, netns_fd);
1337 return system_rtnl_call(msg);
1338 }
1339
1340 static int system_link_del(const char *ifname)
1341 {
1342 struct nl_msg *msg;
1343 struct ifinfomsg iim = {
1344 .ifi_family = AF_UNSPEC,
1345 .ifi_index = 0,
1346 };
1347
1348 msg = nlmsg_alloc_simple(RTM_DELLINK, NLM_F_REQUEST);
1349
1350 if (!msg)
1351 return -1;
1352
1353 nlmsg_append(msg, &iim, sizeof(iim), 0);
1354 nla_put_string(msg, IFLA_IFNAME, ifname);
1355 return system_rtnl_call(msg);
1356 }
1357
1358 int system_macvlan_del(struct device *macvlan)
1359 {
1360 return system_link_del(macvlan->ifname);
1361 }
1362
1363 int system_netns_open(const pid_t target_ns)
1364 {
1365 char pid_net_path[PATH_MAX];
1366
1367 snprintf(pid_net_path, sizeof(pid_net_path), "/proc/%u/ns/net", target_ns);
1368
1369 return open(pid_net_path, O_RDONLY);
1370 }
1371
1372 int system_netns_set(int netns_fd)
1373 {
1374 return setns(netns_fd, CLONE_NEWNET);
1375 }
1376
1377 int system_veth_add(struct device *veth, struct veth_config *cfg)
1378 {
1379 struct nl_msg *msg;
1380 struct ifinfomsg empty_iim = {};
1381 struct nlattr *linkinfo, *data, *veth_info;
1382 int rv;
1383
1384 msg = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_CREATE | NLM_F_EXCL);
1385
1386 if (!msg)
1387 return -1;
1388
1389 nlmsg_append(msg, &empty_iim, sizeof(empty_iim), 0);
1390
1391 if (cfg->flags & VETH_OPT_MACADDR)
1392 nla_put(msg, IFLA_ADDRESS, sizeof(cfg->macaddr), cfg->macaddr);
1393 nla_put_string(msg, IFLA_IFNAME, veth->ifname);
1394
1395 if (!(linkinfo = nla_nest_start(msg, IFLA_LINKINFO)))
1396 goto nla_put_failure;
1397
1398 nla_put_string(msg, IFLA_INFO_KIND, "veth");
1399
1400 if (!(data = nla_nest_start(msg, IFLA_INFO_DATA)))
1401 goto nla_put_failure;
1402
1403 if (!(veth_info = nla_nest_start(msg, VETH_INFO_PEER)))
1404 goto nla_put_failure;
1405
1406 nlmsg_append(msg, &empty_iim, sizeof(empty_iim), 0);
1407
1408 if (cfg->flags & VETH_OPT_PEER_NAME)
1409 nla_put_string(msg, IFLA_IFNAME, cfg->peer_name);
1410 if (cfg->flags & VETH_OPT_PEER_MACADDR)
1411 nla_put(msg, IFLA_ADDRESS, sizeof(cfg->peer_macaddr), cfg->peer_macaddr);
1412
1413 nla_nest_end(msg, veth_info);
1414 nla_nest_end(msg, data);
1415 nla_nest_end(msg, linkinfo);
1416
1417 rv = system_rtnl_call(msg);
1418 if (rv) {
1419 if (cfg->flags & VETH_OPT_PEER_NAME)
1420 D(SYSTEM, "Error adding veth '%s' with peer '%s': %d\n", veth->ifname, cfg->peer_name, rv);
1421 else
1422 D(SYSTEM, "Error adding veth '%s': %d\n", veth->ifname, rv);
1423 }
1424
1425 return rv;
1426
1427 nla_put_failure:
1428 nlmsg_free(msg);
1429 return -ENOMEM;
1430 }
1431
1432 int system_veth_del(struct device *veth)
1433 {
1434 return system_link_del(veth->ifname);
1435 }
1436
1437 static int system_vlan(struct device *dev, int id)
1438 {
1439 struct vlan_ioctl_args ifr = {
1440 .cmd = SET_VLAN_NAME_TYPE_CMD,
1441 .u.name_type = VLAN_NAME_TYPE_RAW_PLUS_VID_NO_PAD,
1442 };
1443
1444 if (ioctl(sock_ioctl, SIOCSIFVLAN, &ifr) < 0)
1445 return -1;
1446
1447 if (id < 0) {
1448 ifr.cmd = DEL_VLAN_CMD;
1449 ifr.u.VID = 0;
1450 } else {
1451 ifr.cmd = ADD_VLAN_CMD;
1452 ifr.u.VID = id;
1453 }
1454 strncpy(ifr.device1, dev->ifname, sizeof(ifr.device1));
1455 return ioctl(sock_ioctl, SIOCSIFVLAN, &ifr);
1456 }
1457
1458 int system_vlan_add(struct device *dev, int id)
1459 {
1460 return system_vlan(dev, id);
1461 }
1462
1463 int system_vlan_del(struct device *dev)
1464 {
1465 return system_vlan(dev, -1);
1466 }
1467
1468 int system_vlandev_add(struct device *vlandev, struct device *dev, struct vlandev_config *cfg)
1469 {
1470 struct nl_msg *msg;
1471 struct nlattr *linkinfo, *data, *qos;
1472 struct ifinfomsg iim = { .ifi_family = AF_UNSPEC };
1473 struct vlan_qos_mapping *elem;
1474 struct ifla_vlan_qos_mapping nl_qos_map;
1475 int rv;
1476
1477 msg = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_CREATE | NLM_F_EXCL);
1478
1479 if (!msg)
1480 return -1;
1481
1482 nlmsg_append(msg, &iim, sizeof(iim), 0);
1483 nla_put_string(msg, IFLA_IFNAME, vlandev->ifname);
1484 nla_put_u32(msg, IFLA_LINK, dev->ifindex);
1485
1486 if (!(linkinfo = nla_nest_start(msg, IFLA_LINKINFO)))
1487 goto nla_put_failure;
1488
1489 nla_put_string(msg, IFLA_INFO_KIND, "vlan");
1490
1491 if (!(data = nla_nest_start(msg, IFLA_INFO_DATA)))
1492 goto nla_put_failure;
1493
1494 nla_put_u16(msg, IFLA_VLAN_ID, cfg->vid);
1495
1496 #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,10,0)
1497 nla_put_u16(msg, IFLA_VLAN_PROTOCOL, htons(cfg->proto));
1498 #else
1499 if(cfg->proto == VLAN_PROTO_8021AD)
1500 netifd_log_message(L_WARNING, "%s Your kernel is older than linux 3.10.0, 802.1ad is not supported defaulting to 802.1q", vlandev->type->name);
1501 #endif
1502
1503 if (!(qos = nla_nest_start(msg, IFLA_VLAN_INGRESS_QOS)))
1504 goto nla_put_failure;
1505
1506 vlist_simple_for_each_element(&cfg->ingress_qos_mapping_list, elem, node) {
1507 nl_qos_map.from = elem->from;
1508 nl_qos_map.to = elem->to;
1509 nla_put(msg, IFLA_VLAN_QOS_MAPPING, sizeof(nl_qos_map), &nl_qos_map);
1510 }
1511 nla_nest_end(msg, qos);
1512
1513 if (!(qos = nla_nest_start(msg, IFLA_VLAN_EGRESS_QOS)))
1514 goto nla_put_failure;
1515
1516 vlist_simple_for_each_element(&cfg->egress_qos_mapping_list, elem, node) {
1517 nl_qos_map.from = elem->from;
1518 nl_qos_map.to = elem->to;
1519 nla_put(msg, IFLA_VLAN_QOS_MAPPING, sizeof(nl_qos_map), &nl_qos_map);
1520 }
1521 nla_nest_end(msg, qos);
1522
1523 nla_nest_end(msg, data);
1524 nla_nest_end(msg, linkinfo);
1525
1526 rv = system_rtnl_call(msg);
1527 if (rv)
1528 D(SYSTEM, "Error adding vlandev '%s' over '%s': %d\n", vlandev->ifname, dev->ifname, rv);
1529
1530 return rv;
1531
1532 nla_put_failure:
1533 nlmsg_free(msg);
1534 return -ENOMEM;
1535 }
1536
1537 int system_vlandev_del(struct device *vlandev)
1538 {
1539 return system_link_del(vlandev->ifname);
1540 }
1541
1542 void
1543 system_if_get_settings(struct device *dev, struct device_settings *s)
1544 {
1545 struct ifreq ifr;
1546 char buf[10];
1547
1548 memset(&ifr, 0, sizeof(ifr));
1549 strncpy(ifr.ifr_name, dev->ifname, sizeof(ifr.ifr_name) - 1);
1550
1551 if (ioctl(sock_ioctl, SIOCGIFMTU, &ifr) == 0) {
1552 s->mtu = ifr.ifr_mtu;
1553 s->flags |= DEV_OPT_MTU;
1554 }
1555
1556 s->mtu6 = system_update_ipv6_mtu(dev, 0);
1557 if (s->mtu6 > 0)
1558 s->flags |= DEV_OPT_MTU6;
1559
1560 if (ioctl(sock_ioctl, SIOCGIFTXQLEN, &ifr) == 0) {
1561 s->txqueuelen = ifr.ifr_qlen;
1562 s->flags |= DEV_OPT_TXQUEUELEN;
1563 }
1564
1565 if (ioctl(sock_ioctl, SIOCGIFHWADDR, &ifr) == 0) {
1566 memcpy(s->macaddr, &ifr.ifr_hwaddr.sa_data, sizeof(s->macaddr));
1567 s->flags |= DEV_OPT_MACADDR;
1568 }
1569
1570 if (!system_get_disable_ipv6(dev, buf, sizeof(buf))) {
1571 s->ipv6 = !strtoul(buf, NULL, 0);
1572 s->flags |= DEV_OPT_IPV6;
1573 }
1574
1575 if (ioctl(sock_ioctl, SIOCGIFFLAGS, &ifr) == 0) {
1576 s->promisc = ifr.ifr_flags & IFF_PROMISC;
1577 s->flags |= DEV_OPT_PROMISC;
1578
1579 s->multicast = ifr.ifr_flags & IFF_MULTICAST;
1580 s->flags |= DEV_OPT_MULTICAST;
1581 }
1582
1583 if (!system_get_rpfilter(dev, buf, sizeof(buf))) {
1584 s->rpfilter = strtoul(buf, NULL, 0);
1585 s->flags |= DEV_OPT_RPFILTER;
1586 }
1587
1588 if (!system_get_acceptlocal(dev, buf, sizeof(buf))) {
1589 s->acceptlocal = strtoul(buf, NULL, 0);
1590 s->flags |= DEV_OPT_ACCEPTLOCAL;
1591 }
1592
1593 if (!system_get_igmpversion(dev, buf, sizeof(buf))) {
1594 s->igmpversion = strtoul(buf, NULL, 0);
1595 s->flags |= DEV_OPT_IGMPVERSION;
1596 }
1597
1598 if (!system_get_mldversion(dev, buf, sizeof(buf))) {
1599 s->mldversion = strtoul(buf, NULL, 0);
1600 s->flags |= DEV_OPT_MLDVERSION;
1601 }
1602
1603 if (!system_get_neigh4reachabletime(dev, buf, sizeof(buf))) {
1604 s->neigh4reachabletime = strtoul(buf, NULL, 0);
1605 s->flags |= DEV_OPT_NEIGHREACHABLETIME;
1606 }
1607
1608 if (!system_get_neigh6reachabletime(dev, buf, sizeof(buf))) {
1609 s->neigh6reachabletime = strtoul(buf, NULL, 0);
1610 s->flags |= DEV_OPT_NEIGHREACHABLETIME;
1611 }
1612
1613 if (!system_get_neigh4locktime(dev, buf, sizeof(buf))) {
1614 s->neigh4locktime = strtol(buf, NULL, 0);
1615 s->flags |= DEV_OPT_NEIGHLOCKTIME;
1616 }
1617
1618 if (!system_get_neigh4gcstaletime(dev, buf, sizeof(buf))) {
1619 s->neigh4gcstaletime = strtoul(buf, NULL, 0);
1620 s->flags |= DEV_OPT_NEIGHGCSTALETIME;
1621 }
1622
1623 if (!system_get_neigh6gcstaletime(dev, buf, sizeof(buf))) {
1624 s->neigh6gcstaletime = strtoul(buf, NULL, 0);
1625 s->flags |= DEV_OPT_NEIGHGCSTALETIME;
1626 }
1627
1628 if (!system_get_dadtransmits(dev, buf, sizeof(buf))) {
1629 s->dadtransmits = strtoul(buf, NULL, 0);
1630 s->flags |= DEV_OPT_DADTRANSMITS;
1631 }
1632
1633 if (!system_get_sendredirects(dev, buf, sizeof(buf))) {
1634 s->sendredirects = strtoul(buf, NULL, 0);
1635 s->flags |= DEV_OPT_SENDREDIRECTS;
1636 }
1637 }
1638
1639 void
1640 system_if_apply_settings(struct device *dev, struct device_settings *s, unsigned int apply_mask)
1641 {
1642 struct ifreq ifr;
1643 char buf[12];
1644
1645 memset(&ifr, 0, sizeof(ifr));
1646 strncpy(ifr.ifr_name, dev->ifname, sizeof(ifr.ifr_name) - 1);
1647 if (s->flags & DEV_OPT_MTU & apply_mask) {
1648 ifr.ifr_mtu = s->mtu;
1649 if (ioctl(sock_ioctl, SIOCSIFMTU, &ifr) < 0)
1650 s->flags &= ~DEV_OPT_MTU;
1651 }
1652 if (s->flags & DEV_OPT_MTU6 & apply_mask) {
1653 system_update_ipv6_mtu(dev, s->mtu6);
1654 }
1655 if (s->flags & DEV_OPT_TXQUEUELEN & apply_mask) {
1656 ifr.ifr_qlen = s->txqueuelen;
1657 if (ioctl(sock_ioctl, SIOCSIFTXQLEN, &ifr) < 0)
1658 s->flags &= ~DEV_OPT_TXQUEUELEN;
1659 }
1660 if ((s->flags & DEV_OPT_MACADDR & apply_mask) && !dev->external) {
1661 ifr.ifr_hwaddr.sa_family = ARPHRD_ETHER;
1662 memcpy(&ifr.ifr_hwaddr.sa_data, s->macaddr, sizeof(s->macaddr));
1663 if (ioctl(sock_ioctl, SIOCSIFHWADDR, &ifr) < 0)
1664 s->flags &= ~DEV_OPT_MACADDR;
1665 }
1666 if (s->flags & DEV_OPT_IPV6 & apply_mask)
1667 system_set_disable_ipv6(dev, s->ipv6 ? "0" : "1");
1668 if (s->flags & DEV_OPT_PROMISC & apply_mask) {
1669 if (system_if_flags(dev->ifname, s->promisc ? IFF_PROMISC : 0,
1670 !s->promisc ? IFF_PROMISC : 0) < 0)
1671 s->flags &= ~DEV_OPT_PROMISC;
1672 }
1673 if (s->flags & DEV_OPT_RPFILTER & apply_mask) {
1674 snprintf(buf, sizeof(buf), "%u", s->rpfilter);
1675 system_set_rpfilter(dev, buf);
1676 }
1677 if (s->flags & DEV_OPT_ACCEPTLOCAL & apply_mask)
1678 system_set_acceptlocal(dev, s->acceptlocal ? "1" : "0");
1679 if (s->flags & DEV_OPT_IGMPVERSION & apply_mask) {
1680 snprintf(buf, sizeof(buf), "%u", s->igmpversion);
1681 system_set_igmpversion(dev, buf);
1682 }
1683 if (s->flags & DEV_OPT_MLDVERSION & apply_mask) {
1684 snprintf(buf, sizeof(buf), "%u", s->mldversion);
1685 system_set_mldversion(dev, buf);
1686 }
1687 if (s->flags & DEV_OPT_NEIGHREACHABLETIME & apply_mask) {
1688 snprintf(buf, sizeof(buf), "%u", s->neigh4reachabletime);
1689 system_set_neigh4reachabletime(dev, buf);
1690 snprintf(buf, sizeof(buf), "%u", s->neigh6reachabletime);
1691 system_set_neigh6reachabletime(dev, buf);
1692 }
1693 if (s->flags & DEV_OPT_NEIGHLOCKTIME & apply_mask) {
1694 snprintf(buf, sizeof(buf), "%d", s->neigh4locktime);
1695 system_set_neigh4locktime(dev, buf);
1696 }
1697 if (s->flags & DEV_OPT_NEIGHGCSTALETIME & apply_mask) {
1698 snprintf(buf, sizeof(buf), "%u", s->neigh4gcstaletime);
1699 system_set_neigh4gcstaletime(dev, buf);
1700 snprintf(buf, sizeof(buf), "%u", s->neigh6gcstaletime);
1701 system_set_neigh6gcstaletime(dev, buf);
1702 }
1703 if (s->flags & DEV_OPT_DADTRANSMITS & apply_mask) {
1704 snprintf(buf, sizeof(buf), "%u", s->dadtransmits);
1705 system_set_dadtransmits(dev, buf);
1706 }
1707 if (s->flags & DEV_OPT_MULTICAST & apply_mask) {
1708 if (system_if_flags(dev->ifname, s->multicast ? IFF_MULTICAST : 0,
1709 !s->multicast ? IFF_MULTICAST : 0) < 0)
1710 s->flags &= ~DEV_OPT_MULTICAST;
1711 }
1712 if (s->flags & DEV_OPT_SENDREDIRECTS & apply_mask)
1713 system_set_sendredirects(dev, s->sendredirects ? "1" : "0");
1714 }
1715
1716 int system_if_up(struct device *dev)
1717 {
1718 system_if_get_settings(dev, &dev->orig_settings);
1719 /* Only keep orig settings based on what needs to be set */
1720 dev->orig_settings.valid_flags = dev->orig_settings.flags;
1721 dev->orig_settings.flags &= dev->settings.flags;
1722 system_if_apply_settings(dev, &dev->settings, dev->settings.flags);
1723 return system_if_flags(dev->ifname, IFF_UP, 0);
1724 }
1725
1726 int system_if_down(struct device *dev)
1727 {
1728 int ret = system_if_flags(dev->ifname, 0, IFF_UP);
1729 system_if_apply_settings(dev, &dev->orig_settings, dev->orig_settings.flags);
1730 return ret;
1731 }
1732
1733 struct if_check_data {
1734 struct device *dev;
1735 int pending;
1736 int ret;
1737 };
1738
1739 #ifndef IFF_LOWER_UP
1740 #define IFF_LOWER_UP 0x10000
1741 #endif
1742
1743 static int cb_if_check_valid(struct nl_msg *msg, void *arg)
1744 {
1745 struct nlmsghdr *nh = nlmsg_hdr(msg);
1746 struct ifinfomsg *ifi = NLMSG_DATA(nh);
1747 struct if_check_data *chk = (struct if_check_data *)arg;
1748
1749 if (nh->nlmsg_type != RTM_NEWLINK)
1750 return NL_SKIP;
1751
1752 if (chk->dev->type == &simple_device_type)
1753 device_set_present(chk->dev, ifi->ifi_index > 0 ? true : false);
1754 device_set_link(chk->dev, ifi->ifi_flags & IFF_LOWER_UP ? true : false);
1755
1756 return NL_OK;
1757 }
1758
1759 static int cb_if_check_ack(struct nl_msg *msg, void *arg)
1760 {
1761 struct if_check_data *chk = (struct if_check_data *)arg;
1762 chk->pending = 0;
1763 return NL_STOP;
1764 }
1765
1766 static int cb_if_check_error(struct sockaddr_nl *nla, struct nlmsgerr *err, void *arg)
1767 {
1768 struct if_check_data *chk = (struct if_check_data *)arg;
1769
1770 if (chk->dev->type == &simple_device_type)
1771 device_set_present(chk->dev, false);
1772 device_set_link(chk->dev, false);
1773 chk->pending = err->error;
1774
1775 return NL_STOP;
1776 }
1777
1778 int system_if_check(struct device *dev)
1779 {
1780 struct nl_cb *cb = nl_cb_alloc(NL_CB_DEFAULT);
1781 struct nl_msg *msg;
1782 struct ifinfomsg ifi = {
1783 .ifi_family = AF_UNSPEC,
1784 .ifi_index = 0,
1785 };
1786 struct if_check_data chk = {
1787 .dev = dev,
1788 .pending = 1,
1789 };
1790 int ret = 1;
1791
1792 if (!cb)
1793 return ret;
1794
1795 msg = nlmsg_alloc_simple(RTM_GETLINK, 0);
1796 if (!msg)
1797 goto out;
1798
1799 if (nlmsg_append(msg, &ifi, sizeof(ifi), 0) ||
1800 nla_put_string(msg, IFLA_IFNAME, dev->ifname))
1801 goto free;
1802
1803 nl_cb_set(cb, NL_CB_VALID, NL_CB_CUSTOM, cb_if_check_valid, &chk);
1804 nl_cb_set(cb, NL_CB_ACK, NL_CB_CUSTOM, cb_if_check_ack, &chk);
1805 nl_cb_err(cb, NL_CB_CUSTOM, cb_if_check_error, &chk);
1806
1807 ret = nl_send_auto_complete(sock_rtnl, msg);
1808 if (ret < 0)
1809 goto free;
1810
1811 while (chk.pending > 0)
1812 nl_recvmsgs(sock_rtnl, cb);
1813
1814 ret = chk.pending;
1815
1816 free:
1817 nlmsg_free(msg);
1818 out:
1819 nl_cb_put(cb);
1820 return ret;
1821 }
1822
1823 struct device *
1824 system_if_get_parent(struct device *dev)
1825 {
1826 char buf[64], *devname;
1827 int ifindex, iflink, len;
1828 FILE *f;
1829
1830 snprintf(buf, sizeof(buf), "/sys/class/net/%s/iflink", dev->ifname);
1831 f = fopen(buf, "r");
1832 if (!f)
1833 return NULL;
1834
1835 len = fread(buf, 1, sizeof(buf) - 1, f);
1836 fclose(f);
1837
1838 if (len <= 0)
1839 return NULL;
1840
1841 buf[len] = 0;
1842 iflink = strtoul(buf, NULL, 0);
1843 ifindex = system_if_resolve(dev);
1844 if (!iflink || iflink == ifindex)
1845 return NULL;
1846
1847 devname = if_indextoname(iflink, buf);
1848 if (!devname)
1849 return NULL;
1850
1851 return device_get(devname, true);
1852 }
1853
1854 static bool
1855 read_string_file(int dir_fd, const char *file, char *buf, int len)
1856 {
1857 bool ret = false;
1858 char *c;
1859 int fd;
1860
1861 fd = openat(dir_fd, file, O_RDONLY);
1862 if (fd < 0)
1863 return false;
1864
1865 retry:
1866 len = read(fd, buf, len - 1);
1867 if (len < 0) {
1868 if (errno == EINTR)
1869 goto retry;
1870 } else if (len > 0) {
1871 buf[len] = 0;
1872
1873 c = strchr(buf, '\n');
1874 if (c)
1875 *c = 0;
1876
1877 ret = true;
1878 }
1879
1880 close(fd);
1881
1882 return ret;
1883 }
1884
1885 static bool
1886 read_uint64_file(int dir_fd, const char *file, uint64_t *val)
1887 {
1888 char buf[64];
1889 bool ret = false;
1890
1891 ret = read_string_file(dir_fd, file, buf, sizeof(buf));
1892 if (ret)
1893 *val = strtoull(buf, NULL, 0);
1894
1895 return ret;
1896 }
1897
1898 /* Assume advertised flags == supported flags */
1899 static const struct {
1900 uint32_t mask;
1901 const char *name;
1902 } ethtool_link_modes[] = {
1903 { ADVERTISED_10baseT_Half, "10baseT-H" },
1904 { ADVERTISED_10baseT_Full, "10baseT-F" },
1905 { ADVERTISED_100baseT_Half, "100baseT-H" },
1906 { ADVERTISED_100baseT_Full, "100baseT-F" },
1907 { ADVERTISED_1000baseT_Half, "1000baseT-H" },
1908 { ADVERTISED_1000baseT_Full, "1000baseT-F" },
1909 { ADVERTISED_1000baseKX_Full, "1000baseKX-F" },
1910 { ADVERTISED_2500baseX_Full, "2500baseX-F" },
1911 { ADVERTISED_10000baseT_Full, "10000baseT-F" },
1912 { ADVERTISED_10000baseKX4_Full, "10000baseKX4-F" },
1913 { ADVERTISED_10000baseKR_Full, "10000baseKR-F" },
1914 { ADVERTISED_20000baseMLD2_Full, "20000baseMLD2-F" },
1915 { ADVERTISED_20000baseKR2_Full, "20000baseKR2-F" },
1916 { ADVERTISED_40000baseKR4_Full, "40000baseKR4-F" },
1917 { ADVERTISED_40000baseCR4_Full, "40000baseCR4-F" },
1918 { ADVERTISED_40000baseSR4_Full, "40000baseSR4-F" },
1919 { ADVERTISED_40000baseLR4_Full, "40000baseLR4-F" },
1920 #ifdef ADVERTISED_56000baseKR4_Full
1921 { ADVERTISED_56000baseKR4_Full, "56000baseKR4-F" },
1922 { ADVERTISED_56000baseCR4_Full, "56000baseCR4-F" },
1923 { ADVERTISED_56000baseSR4_Full, "56000baseSR4-F" },
1924 { ADVERTISED_56000baseLR4_Full, "56000baseLR4-F" },
1925 #endif
1926 };
1927
1928 static void system_add_link_modes(struct blob_buf *b, __u32 mask)
1929 {
1930 int i;
1931 for (i = 0; i < ARRAY_SIZE(ethtool_link_modes); i++) {
1932 if (mask & ethtool_link_modes[i].mask)
1933 blobmsg_add_string(b, NULL, ethtool_link_modes[i].name);
1934 }
1935 }
1936
1937 bool
1938 system_if_force_external(const char *ifname)
1939 {
1940 char buf[64];
1941 struct stat s;
1942
1943 snprintf(buf, sizeof(buf), "/sys/class/net/%s/phy80211", ifname);
1944 return stat(buf, &s) == 0;
1945 }
1946
1947 int
1948 system_if_dump_info(struct device *dev, struct blob_buf *b)
1949 {
1950 struct ethtool_cmd ecmd;
1951 struct ifreq ifr;
1952 char *s;
1953 void *c;
1954
1955 memset(&ecmd, 0, sizeof(ecmd));
1956 memset(&ifr, 0, sizeof(ifr));
1957 strncpy(ifr.ifr_name, dev->ifname, sizeof(ifr.ifr_name) - 1);
1958 ifr.ifr_data = (caddr_t) &ecmd;
1959 ecmd.cmd = ETHTOOL_GSET;
1960
1961 if (ioctl(sock_ioctl, SIOCETHTOOL, &ifr) == 0) {
1962 c = blobmsg_open_array(b, "link-advertising");
1963 system_add_link_modes(b, ecmd.advertising);
1964 blobmsg_close_array(b, c);
1965
1966 c = blobmsg_open_array(b, "link-partner-advertising");
1967 system_add_link_modes(b, ecmd.lp_advertising);
1968 blobmsg_close_array(b, c);
1969
1970 c = blobmsg_open_array(b, "link-supported");
1971 system_add_link_modes(b, ecmd.supported);
1972 blobmsg_close_array(b, c);
1973
1974 s = blobmsg_alloc_string_buffer(b, "speed", 8);
1975 snprintf(s, 8, "%d%c", ethtool_cmd_speed(&ecmd),
1976 ecmd.duplex == DUPLEX_HALF ? 'H' : 'F');
1977 blobmsg_add_string_buffer(b);
1978
1979 blobmsg_add_u8(b, "autoneg", !!ecmd.autoneg);
1980 }
1981
1982 return 0;
1983 }
1984
1985 int
1986 system_if_dump_stats(struct device *dev, struct blob_buf *b)
1987 {
1988 const char *const counters[] = {
1989 "collisions", "rx_frame_errors", "tx_compressed",
1990 "multicast", "rx_length_errors", "tx_dropped",
1991 "rx_bytes", "rx_missed_errors", "tx_errors",
1992 "rx_compressed", "rx_over_errors", "tx_fifo_errors",
1993 "rx_crc_errors", "rx_packets", "tx_heartbeat_errors",
1994 "rx_dropped", "tx_aborted_errors", "tx_packets",
1995 "rx_errors", "tx_bytes", "tx_window_errors",
1996 "rx_fifo_errors", "tx_carrier_errors",
1997 };
1998 char buf[64];
1999 int stats_dir;
2000 int i;
2001 uint64_t val = 0;
2002
2003 snprintf(buf, sizeof(buf), "/sys/class/net/%s/statistics", dev->ifname);
2004 stats_dir = open(buf, O_DIRECTORY);
2005 if (stats_dir < 0)
2006 return -1;
2007
2008 for (i = 0; i < ARRAY_SIZE(counters); i++)
2009 if (read_uint64_file(stats_dir, counters[i], &val))
2010 blobmsg_add_u64(b, counters[i], val);
2011
2012 close(stats_dir);
2013 return 0;
2014 }
2015
2016 static int system_addr(struct device *dev, struct device_addr *addr, int cmd)
2017 {
2018 bool v4 = ((addr->flags & DEVADDR_FAMILY) == DEVADDR_INET4);
2019 int alen = v4 ? 4 : 16;
2020 unsigned int flags = 0;
2021 struct ifaddrmsg ifa = {
2022 .ifa_family = (alen == 4) ? AF_INET : AF_INET6,
2023 .ifa_prefixlen = addr->mask,
2024 .ifa_index = dev->ifindex,
2025 };
2026
2027 struct nl_msg *msg;
2028 if (cmd == RTM_NEWADDR)
2029 flags |= NLM_F_CREATE | NLM_F_REPLACE;
2030
2031 msg = nlmsg_alloc_simple(cmd, flags);
2032 if (!msg)
2033 return -1;
2034
2035 nlmsg_append(msg, &ifa, sizeof(ifa), 0);
2036 nla_put(msg, IFA_LOCAL, alen, &addr->addr);
2037 if (v4) {
2038 if (addr->broadcast)
2039 nla_put_u32(msg, IFA_BROADCAST, addr->broadcast);
2040 if (addr->point_to_point)
2041 nla_put_u32(msg, IFA_ADDRESS, addr->point_to_point);
2042 } else {
2043 time_t now = system_get_rtime();
2044 struct ifa_cacheinfo cinfo = {0xffffffffU, 0xffffffffU, 0, 0};
2045
2046 if (addr->preferred_until) {
2047 int64_t preferred = addr->preferred_until - now;
2048 if (preferred < 0)
2049 preferred = 0;
2050 else if (preferred > UINT32_MAX)
2051 preferred = UINT32_MAX;
2052
2053 cinfo.ifa_prefered = preferred;
2054 }
2055
2056 if (addr->valid_until) {
2057 int64_t valid = addr->valid_until - now;
2058 if (valid <= 0) {
2059 nlmsg_free(msg);
2060 return -1;
2061 }
2062 else if (valid > UINT32_MAX)
2063 valid = UINT32_MAX;
2064
2065 cinfo.ifa_valid = valid;
2066 }
2067
2068 nla_put(msg, IFA_CACHEINFO, sizeof(cinfo), &cinfo);
2069
2070 if (cmd == RTM_NEWADDR && (addr->flags & DEVADDR_OFFLINK))
2071 nla_put_u32(msg, IFA_FLAGS, IFA_F_NOPREFIXROUTE);
2072 }
2073
2074 return system_rtnl_call(msg);
2075 }
2076
2077 int system_add_address(struct device *dev, struct device_addr *addr)
2078 {
2079 return system_addr(dev, addr, RTM_NEWADDR);
2080 }
2081
2082 int system_del_address(struct device *dev, struct device_addr *addr)
2083 {
2084 return system_addr(dev, addr, RTM_DELADDR);
2085 }
2086
2087 static int system_neigh(struct device *dev, struct device_neighbor *neighbor, int cmd)
2088 {
2089 int alen = ((neighbor->flags & DEVADDR_FAMILY) == DEVADDR_INET4) ? 4 : 16;
2090 unsigned int flags = 0;
2091 struct ndmsg ndm = {
2092 .ndm_family = (alen == 4) ? AF_INET : AF_INET6,
2093 .ndm_ifindex = dev->ifindex,
2094 .ndm_state = NUD_PERMANENT,
2095 .ndm_flags = (neighbor->proxy ? NTF_PROXY : 0) | (neighbor->router ? NTF_ROUTER : 0),
2096 };
2097 struct nl_msg *msg;
2098
2099 if (cmd == RTM_NEWNEIGH)
2100 flags |= NLM_F_CREATE | NLM_F_REPLACE;
2101
2102 msg = nlmsg_alloc_simple(cmd, flags);
2103
2104 if (!msg)
2105 return -1;
2106
2107 nlmsg_append(msg, &ndm, sizeof(ndm), 0);
2108
2109 nla_put(msg, NDA_DST, alen, &neighbor->addr);
2110 if (neighbor->flags & DEVNEIGH_MAC)
2111 nla_put(msg, NDA_LLADDR, sizeof(neighbor->macaddr), &neighbor->macaddr);
2112
2113
2114 return system_rtnl_call(msg);
2115 }
2116
2117 int system_add_neighbor(struct device *dev, struct device_neighbor *neighbor)
2118 {
2119 return system_neigh(dev, neighbor, RTM_NEWNEIGH);
2120 }
2121
2122 int system_del_neighbor(struct device *dev, struct device_neighbor *neighbor)
2123 {
2124 return system_neigh(dev, neighbor, RTM_DELNEIGH);
2125 }
2126
2127 static int system_rt(struct device *dev, struct device_route *route, int cmd)
2128 {
2129 int alen = ((route->flags & DEVADDR_FAMILY) == DEVADDR_INET4) ? 4 : 16;
2130 bool have_gw;
2131 unsigned int flags = 0;
2132
2133 if (alen == 4)
2134 have_gw = !!route->nexthop.in.s_addr;
2135 else
2136 have_gw = route->nexthop.in6.s6_addr32[0] ||
2137 route->nexthop.in6.s6_addr32[1] ||
2138 route->nexthop.in6.s6_addr32[2] ||
2139 route->nexthop.in6.s6_addr32[3];
2140
2141 unsigned int table = (route->flags & (DEVROUTE_TABLE | DEVROUTE_SRCTABLE))
2142 ? route->table : RT_TABLE_MAIN;
2143
2144 struct rtmsg rtm = {
2145 .rtm_family = (alen == 4) ? AF_INET : AF_INET6,
2146 .rtm_dst_len = route->mask,
2147 .rtm_src_len = route->sourcemask,
2148 .rtm_table = (table < 256) ? table : RT_TABLE_UNSPEC,
2149 .rtm_protocol = (route->flags & DEVROUTE_PROTO) ? route->proto : RTPROT_STATIC,
2150 .rtm_scope = RT_SCOPE_NOWHERE,
2151 .rtm_type = (cmd == RTM_DELROUTE) ? 0: RTN_UNICAST,
2152 .rtm_flags = (route->flags & DEVROUTE_ONLINK) ? RTNH_F_ONLINK : 0,
2153 };
2154 struct nl_msg *msg;
2155
2156 if (cmd == RTM_NEWROUTE) {
2157 flags |= NLM_F_CREATE | NLM_F_REPLACE;
2158
2159 if (!dev) { /* Add null-route */
2160 rtm.rtm_scope = RT_SCOPE_UNIVERSE;
2161 rtm.rtm_type = RTN_UNREACHABLE;
2162 }
2163 else
2164 rtm.rtm_scope = (have_gw) ? RT_SCOPE_UNIVERSE : RT_SCOPE_LINK;
2165 }
2166
2167 if (route->flags & DEVROUTE_TYPE) {
2168 rtm.rtm_type = route->type;
2169 if (!(route->flags & (DEVROUTE_TABLE | DEVROUTE_SRCTABLE))) {
2170 if (rtm.rtm_type == RTN_LOCAL || rtm.rtm_type == RTN_BROADCAST ||
2171 rtm.rtm_type == RTN_NAT || rtm.rtm_type == RTN_ANYCAST)
2172 rtm.rtm_table = RT_TABLE_LOCAL;
2173 }
2174
2175 if (rtm.rtm_type == RTN_LOCAL || rtm.rtm_type == RTN_NAT) {
2176 rtm.rtm_scope = RT_SCOPE_HOST;
2177 } else if (rtm.rtm_type == RTN_BROADCAST || rtm.rtm_type == RTN_MULTICAST ||
2178 rtm.rtm_type == RTN_ANYCAST) {
2179 rtm.rtm_scope = RT_SCOPE_LINK;
2180 } else if (rtm.rtm_type == RTN_BLACKHOLE || rtm.rtm_type == RTN_UNREACHABLE ||
2181 rtm.rtm_type == RTN_PROHIBIT || rtm.rtm_type == RTN_FAILED_POLICY ||
2182 rtm.rtm_type == RTN_THROW) {
2183 rtm.rtm_scope = RT_SCOPE_UNIVERSE;
2184 dev = NULL;
2185 }
2186 }
2187
2188 msg = nlmsg_alloc_simple(cmd, flags);
2189 if (!msg)
2190 return -1;
2191
2192 nlmsg_append(msg, &rtm, sizeof(rtm), 0);
2193
2194 if (route->mask)
2195 nla_put(msg, RTA_DST, alen, &route->addr);
2196
2197 if (route->sourcemask) {
2198 if (rtm.rtm_family == AF_INET)
2199 nla_put(msg, RTA_PREFSRC, alen, &route->source);
2200 else
2201 nla_put(msg, RTA_SRC, alen, &route->source);
2202 }
2203
2204 if (route->metric > 0)
2205 nla_put_u32(msg, RTA_PRIORITY, route->metric);
2206
2207 if (have_gw)
2208 nla_put(msg, RTA_GATEWAY, alen, &route->nexthop);
2209
2210 if (dev)
2211 nla_put_u32(msg, RTA_OIF, dev->ifindex);
2212
2213 if (table >= 256)
2214 nla_put_u32(msg, RTA_TABLE, table);
2215
2216 if (route->flags & DEVROUTE_MTU) {
2217 struct nlattr *metrics;
2218
2219 if (!(metrics = nla_nest_start(msg, RTA_METRICS)))
2220 goto nla_put_failure;
2221
2222 nla_put_u32(msg, RTAX_MTU, route->mtu);
2223
2224 nla_nest_end(msg, metrics);
2225 }
2226
2227 return system_rtnl_call(msg);
2228
2229 nla_put_failure:
2230 nlmsg_free(msg);
2231 return -ENOMEM;
2232 }
2233
2234 int system_add_route(struct device *dev, struct device_route *route)
2235 {
2236 return system_rt(dev, route, RTM_NEWROUTE);
2237 }
2238
2239 int system_del_route(struct device *dev, struct device_route *route)
2240 {
2241 return system_rt(dev, route, RTM_DELROUTE);
2242 }
2243
2244 int system_flush_routes(void)
2245 {
2246 const char *names[] = {
2247 "/proc/sys/net/ipv4/route/flush",
2248 "/proc/sys/net/ipv6/route/flush"
2249 };
2250 int fd, i;
2251
2252 for (i = 0; i < ARRAY_SIZE(names); i++) {
2253 fd = open(names[i], O_WRONLY);
2254 if (fd < 0)
2255 continue;
2256
2257 if (write(fd, "-1", 2)) {}
2258 close(fd);
2259 }
2260 return 0;
2261 }
2262
2263 bool system_resolve_rt_type(const char *type, unsigned int *id)
2264 {
2265 return system_rtn_aton(type, id);
2266 }
2267
2268 bool system_resolve_rt_proto(const char *type, unsigned int *id)
2269 {
2270 FILE *f;
2271 char *e, buf[128];
2272 unsigned int n, proto = 256;
2273 n = strtoul(type, &e, 0);
2274 if (!*e && e != type)
2275 proto = n;
2276 else if (!strcmp(type, "unspec"))
2277 proto = RTPROT_UNSPEC;
2278 else if (!strcmp(type, "kernel"))
2279 proto = RTPROT_KERNEL;
2280 else if (!strcmp(type, "boot"))
2281 proto = RTPROT_BOOT;
2282 else if (!strcmp(type, "static"))
2283 proto = RTPROT_STATIC;
2284 else if ((f = fopen("/etc/iproute2/rt_protos", "r")) != NULL) {
2285 while (fgets(buf, sizeof(buf) - 1, f) != NULL) {
2286 if ((e = strtok(buf, " \t\n")) == NULL || *e == '#')
2287 continue;
2288
2289 n = strtoul(e, NULL, 10);
2290 e = strtok(NULL, " \t\n");
2291
2292 if (e && !strcmp(e, type)) {
2293 proto = n;
2294 break;
2295 }
2296 }
2297 fclose(f);
2298 }
2299
2300 if (proto > 255)
2301 return false;
2302
2303 *id = proto;
2304 return true;
2305 }
2306
2307 bool system_resolve_rt_table(const char *name, unsigned int *id)
2308 {
2309 FILE *f;
2310 char *e, buf[128];
2311 unsigned int n, table = RT_TABLE_UNSPEC;
2312
2313 /* first try to parse table as number */
2314 if ((n = strtoul(name, &e, 0)) > 0 && !*e)
2315 table = n;
2316
2317 /* handle well known aliases */
2318 else if (!strcmp(name, "default"))
2319 table = RT_TABLE_DEFAULT;
2320 else if (!strcmp(name, "main"))
2321 table = RT_TABLE_MAIN;
2322 else if (!strcmp(name, "local"))
2323 table = RT_TABLE_LOCAL;
2324
2325 /* try to look up name in /etc/iproute2/rt_tables */
2326 else if ((f = fopen("/etc/iproute2/rt_tables", "r")) != NULL)
2327 {
2328 while (fgets(buf, sizeof(buf) - 1, f) != NULL)
2329 {
2330 if ((e = strtok(buf, " \t\n")) == NULL || *e == '#')
2331 continue;
2332
2333 n = strtoul(e, NULL, 10);
2334 e = strtok(NULL, " \t\n");
2335
2336 if (e && !strcmp(e, name))
2337 {
2338 table = n;
2339 break;
2340 }
2341 }
2342
2343 fclose(f);
2344 }
2345
2346 if (table == RT_TABLE_UNSPEC)
2347 return false;
2348
2349 *id = table;
2350 return true;
2351 }
2352
2353 bool system_is_default_rt_table(unsigned int id)
2354 {
2355 return (id == RT_TABLE_MAIN);
2356 }
2357
2358 bool system_resolve_rpfilter(const char *filter, unsigned int *id)
2359 {
2360 char *e;
2361 unsigned int n;
2362
2363 if (!strcmp(filter, "strict"))
2364 n = 1;
2365 else if (!strcmp(filter, "loose"))
2366 n = 2;
2367 else {
2368 n = strtoul(filter, &e, 0);
2369 if (*e || e == filter || n > 2)
2370 return false;
2371 }
2372
2373 *id = n;
2374 return true;
2375 }
2376
2377 static int system_iprule(struct iprule *rule, int cmd)
2378 {
2379 int alen = ((rule->flags & IPRULE_FAMILY) == IPRULE_INET4) ? 4 : 16;
2380
2381 struct nl_msg *msg;
2382 struct rtmsg rtm = {
2383 .rtm_family = (alen == 4) ? AF_INET : AF_INET6,
2384 .rtm_protocol = RTPROT_STATIC,
2385 .rtm_scope = RT_SCOPE_UNIVERSE,
2386 .rtm_table = RT_TABLE_UNSPEC,
2387 .rtm_type = RTN_UNSPEC,
2388 .rtm_flags = 0,
2389 };
2390
2391 if (cmd == RTM_NEWRULE)
2392 rtm.rtm_type = RTN_UNICAST;
2393
2394 if (rule->invert)
2395 rtm.rtm_flags |= FIB_RULE_INVERT;
2396
2397 if (rule->flags & IPRULE_SRC)
2398 rtm.rtm_src_len = rule->src_mask;
2399
2400 if (rule->flags & IPRULE_DEST)
2401 rtm.rtm_dst_len = rule->dest_mask;
2402
2403 if (rule->flags & IPRULE_TOS)
2404 rtm.rtm_tos = rule->tos;
2405
2406 if (rule->flags & IPRULE_LOOKUP) {
2407 if (rule->lookup < 256)
2408 rtm.rtm_table = rule->lookup;
2409 }
2410
2411 if (rule->flags & IPRULE_ACTION)
2412 rtm.rtm_type = rule->action;
2413 else if (rule->flags & IPRULE_GOTO)
2414 rtm.rtm_type = FR_ACT_GOTO;
2415 else if (!(rule->flags & (IPRULE_LOOKUP | IPRULE_ACTION | IPRULE_GOTO)))
2416 rtm.rtm_type = FR_ACT_NOP;
2417
2418 msg = nlmsg_alloc_simple(cmd, NLM_F_REQUEST);
2419
2420 if (!msg)
2421 return -1;
2422
2423 nlmsg_append(msg, &rtm, sizeof(rtm), 0);
2424
2425 if (rule->flags & IPRULE_IN)
2426 nla_put(msg, FRA_IFNAME, strlen(rule->in_dev) + 1, rule->in_dev);
2427
2428 if (rule->flags & IPRULE_OUT)
2429 nla_put(msg, FRA_OIFNAME, strlen(rule->out_dev) + 1, rule->out_dev);
2430
2431 if (rule->flags & IPRULE_SRC)
2432 nla_put(msg, FRA_SRC, alen, &rule->src_addr);
2433
2434 if (rule->flags & IPRULE_DEST)
2435 nla_put(msg, FRA_DST, alen, &rule->dest_addr);
2436
2437 if (rule->flags & IPRULE_PRIORITY)
2438 nla_put_u32(msg, FRA_PRIORITY, rule->priority);
2439 else if (cmd == RTM_NEWRULE)
2440 nla_put_u32(msg, FRA_PRIORITY, rule->order);
2441
2442 if (rule->flags & IPRULE_FWMARK)
2443 nla_put_u32(msg, FRA_FWMARK, rule->fwmark);
2444
2445 if (rule->flags & IPRULE_FWMASK)
2446 nla_put_u32(msg, FRA_FWMASK, rule->fwmask);
2447
2448 if (rule->flags & IPRULE_LOOKUP) {
2449 if (rule->lookup >= 256)
2450 nla_put_u32(msg, FRA_TABLE, rule->lookup);
2451 }
2452
2453 if (rule->flags & IPRULE_SUP_PREFIXLEN)
2454 nla_put_u32(msg, FRA_SUPPRESS_PREFIXLEN, rule->sup_prefixlen);
2455
2456 if (rule->flags & IPRULE_GOTO)
2457 nla_put_u32(msg, FRA_GOTO, rule->gotoid);
2458
2459 return system_rtnl_call(msg);
2460 }
2461
2462 int system_add_iprule(struct iprule *rule)
2463 {
2464 return system_iprule(rule, RTM_NEWRULE);
2465 }
2466
2467 int system_del_iprule(struct iprule *rule)
2468 {
2469 return system_iprule(rule, RTM_DELRULE);
2470 }
2471
2472 int system_flush_iprules(void)
2473 {
2474 int rv = 0;
2475 struct iprule rule;
2476
2477 system_if_clear_entries(NULL, RTM_GETRULE, AF_INET);
2478 system_if_clear_entries(NULL, RTM_GETRULE, AF_INET6);
2479
2480 memset(&rule, 0, sizeof(rule));
2481
2482
2483 rule.flags = IPRULE_INET4 | IPRULE_PRIORITY | IPRULE_LOOKUP;
2484
2485 rule.priority = 0;
2486 rule.lookup = RT_TABLE_LOCAL;
2487 rv |= system_iprule(&rule, RTM_NEWRULE);
2488
2489 rule.priority = 32766;
2490 rule.lookup = RT_TABLE_MAIN;
2491 rv |= system_iprule(&rule, RTM_NEWRULE);
2492
2493 rule.priority = 32767;
2494 rule.lookup = RT_TABLE_DEFAULT;
2495 rv |= system_iprule(&rule, RTM_NEWRULE);
2496
2497
2498 rule.flags = IPRULE_INET6 | IPRULE_PRIORITY | IPRULE_LOOKUP;
2499
2500 rule.priority = 0;
2501 rule.lookup = RT_TABLE_LOCAL;
2502 rv |= system_iprule(&rule, RTM_NEWRULE);
2503
2504 rule.priority = 32766;
2505 rule.lookup = RT_TABLE_MAIN;
2506 rv |= system_iprule(&rule, RTM_NEWRULE);
2507
2508 return rv;
2509 }
2510
2511 bool system_resolve_iprule_action(const char *action, unsigned int *id)
2512 {
2513 return system_rtn_aton(action, id);
2514 }
2515
2516 time_t system_get_rtime(void)
2517 {
2518 struct timespec ts;
2519 struct timeval tv;
2520
2521 if (clock_gettime(CLOCK_MONOTONIC, &ts) == 0)
2522 return ts.tv_sec;
2523
2524 if (gettimeofday(&tv, NULL) == 0)
2525 return tv.tv_sec;
2526
2527 return 0;
2528 }
2529
2530 #ifndef IP_DF
2531 #define IP_DF 0x4000
2532 #endif
2533
2534 static int tunnel_ioctl(const char *name, int cmd, void *p)
2535 {
2536 struct ifreq ifr;
2537
2538 memset(&ifr, 0, sizeof(ifr));
2539 strncpy(ifr.ifr_name, name, sizeof(ifr.ifr_name) - 1);
2540 ifr.ifr_ifru.ifru_data = p;
2541 return ioctl(sock_ioctl, cmd, &ifr);
2542 }
2543
2544 #ifdef IFLA_IPTUN_MAX
2545 static int system_add_ip6_tunnel(const char *name, const unsigned int link,
2546 struct blob_attr **tb)
2547 {
2548 struct nl_msg *nlm = nlmsg_alloc_simple(RTM_NEWLINK,
2549 NLM_F_REQUEST | NLM_F_REPLACE | NLM_F_CREATE);
2550 struct ifinfomsg ifi = { .ifi_family = AF_UNSPEC };
2551 struct blob_attr *cur;
2552 int ret = 0, ttl = 0;
2553
2554 if (!nlm)
2555 return -1;
2556
2557 nlmsg_append(nlm, &ifi, sizeof(ifi), 0);
2558 nla_put_string(nlm, IFLA_IFNAME, name);
2559
2560 if (link)
2561 nla_put_u32(nlm, IFLA_LINK, link);
2562
2563 struct nlattr *linkinfo = nla_nest_start(nlm, IFLA_LINKINFO);
2564 if (!linkinfo) {
2565 ret = -ENOMEM;
2566 goto failure;
2567 }
2568
2569 nla_put_string(nlm, IFLA_INFO_KIND, "ip6tnl");
2570 struct nlattr *infodata = nla_nest_start(nlm, IFLA_INFO_DATA);
2571 if (!infodata) {
2572 ret = -ENOMEM;
2573 goto failure;
2574 }
2575
2576 if (link)
2577 nla_put_u32(nlm, IFLA_IPTUN_LINK, link);
2578
2579 if ((cur = tb[TUNNEL_ATTR_TTL]))
2580 ttl = blobmsg_get_u32(cur);
2581
2582 nla_put_u8(nlm, IFLA_IPTUN_PROTO, IPPROTO_IPIP);
2583 nla_put_u8(nlm, IFLA_IPTUN_TTL, (ttl) ? ttl : 64);
2584
2585 struct in6_addr in6buf;
2586 if ((cur = tb[TUNNEL_ATTR_LOCAL])) {
2587 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
2588 ret = -EINVAL;
2589 goto failure;
2590 }
2591 nla_put(nlm, IFLA_IPTUN_LOCAL, sizeof(in6buf), &in6buf);
2592 }
2593
2594 if ((cur = tb[TUNNEL_ATTR_REMOTE])) {
2595 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
2596 ret = -EINVAL;
2597 goto failure;
2598 }
2599 nla_put(nlm, IFLA_IPTUN_REMOTE, sizeof(in6buf), &in6buf);
2600 }
2601
2602 if ((cur = tb[TUNNEL_ATTR_DATA])) {
2603 struct blob_attr *tb_data[__IPIP6_DATA_ATTR_MAX];
2604 uint32_t tun_flags = IP6_TNL_F_IGN_ENCAP_LIMIT;
2605
2606 blobmsg_parse(ipip6_data_attr_list.params, __IPIP6_DATA_ATTR_MAX, tb_data,
2607 blobmsg_data(cur), blobmsg_len(cur));
2608
2609 if ((cur = tb_data[IPIP6_DATA_ENCAPLIMIT])) {
2610 char *str = blobmsg_get_string(cur);
2611
2612 if (strcmp(str, "ignore")) {
2613 char *e;
2614 unsigned encap_limit = strtoul(str, &e, 0);
2615
2616 if (e == str || *e || encap_limit > 255) {
2617 ret = -EINVAL;
2618 goto failure;
2619 }
2620
2621 nla_put_u8(nlm, IFLA_IPTUN_ENCAP_LIMIT, encap_limit);
2622 tun_flags &= ~IP6_TNL_F_IGN_ENCAP_LIMIT;
2623 }
2624 }
2625
2626 #ifdef IFLA_IPTUN_FMR_MAX
2627 if ((cur = tb_data[IPIP6_DATA_FMRS])) {
2628 struct blob_attr *rcur;
2629 unsigned rrem, fmrcnt = 0;
2630 struct nlattr *fmrs = nla_nest_start(nlm, IFLA_IPTUN_FMRS);
2631
2632 if (!fmrs) {
2633 ret = -ENOMEM;
2634 goto failure;
2635 }
2636
2637 blobmsg_for_each_attr(rcur, cur, rrem) {
2638 struct blob_attr *tb_fmr[__FMR_DATA_ATTR_MAX], *tb_cur;
2639 struct in6_addr ip6prefix;
2640 struct in_addr ip4prefix;
2641 unsigned ip4len, ip6len, ealen, offset;
2642
2643 blobmsg_parse(fmr_data_attr_list.params, __FMR_DATA_ATTR_MAX, tb_fmr,
2644 blobmsg_data(rcur), blobmsg_len(rcur));
2645
2646 if (!(tb_cur = tb_fmr[FMR_DATA_PREFIX6]) ||
2647 !parse_ip_and_netmask(AF_INET6,
2648 blobmsg_data(tb_cur), &ip6prefix,
2649 &ip6len)) {
2650 ret = -EINVAL;
2651 goto failure;
2652 }
2653
2654 if (!(tb_cur = tb_fmr[FMR_DATA_PREFIX4]) ||
2655 !parse_ip_and_netmask(AF_INET,
2656 blobmsg_data(tb_cur), &ip4prefix,
2657 &ip4len)) {
2658 ret = -EINVAL;
2659 goto failure;
2660 }
2661
2662 if (!(tb_cur = tb_fmr[FMR_DATA_EALEN])) {
2663 ret = -EINVAL;
2664 goto failure;
2665 }
2666 ealen = blobmsg_get_u32(tb_cur);
2667
2668 if (!(tb_cur = tb_fmr[FMR_DATA_OFFSET])) {
2669 ret = -EINVAL;
2670 goto failure;
2671 }
2672 offset = blobmsg_get_u32(tb_cur);
2673
2674 struct nlattr *rule = nla_nest_start(nlm, ++fmrcnt);
2675 if (!rule) {
2676 ret = -ENOMEM;
2677 goto failure;
2678 }
2679
2680 nla_put(nlm, IFLA_IPTUN_FMR_IP6_PREFIX, sizeof(ip6prefix), &ip6prefix);
2681 nla_put(nlm, IFLA_IPTUN_FMR_IP4_PREFIX, sizeof(ip4prefix), &ip4prefix);
2682 nla_put_u8(nlm, IFLA_IPTUN_FMR_IP6_PREFIX_LEN, ip6len);
2683 nla_put_u8(nlm, IFLA_IPTUN_FMR_IP4_PREFIX_LEN, ip4len);
2684 nla_put_u8(nlm, IFLA_IPTUN_FMR_EA_LEN, ealen);
2685 nla_put_u8(nlm, IFLA_IPTUN_FMR_OFFSET, offset);
2686
2687 nla_nest_end(nlm, rule);
2688 }
2689
2690 nla_nest_end(nlm, fmrs);
2691 }
2692 #endif
2693 if (tun_flags)
2694 nla_put_u32(nlm, IFLA_IPTUN_FLAGS, tun_flags);
2695 }
2696
2697 nla_nest_end(nlm, infodata);
2698 nla_nest_end(nlm, linkinfo);
2699
2700 return system_rtnl_call(nlm);
2701
2702 failure:
2703 nlmsg_free(nlm);
2704 return ret;
2705 }
2706 #endif
2707
2708 #ifdef IFLA_IPTUN_MAX
2709 #define IP6_FLOWINFO_TCLASS htonl(0x0FF00000)
2710 static int system_add_gre_tunnel(const char *name, const char *kind,
2711 const unsigned int link, struct blob_attr **tb, bool v6)
2712 {
2713 struct nl_msg *nlm;
2714 struct ifinfomsg ifi = { .ifi_family = AF_UNSPEC, };
2715 struct blob_attr *cur;
2716 uint32_t ikey = 0, okey = 0, flowinfo = 0, flags6 = IP6_TNL_F_IGN_ENCAP_LIMIT;
2717 uint16_t iflags = 0, oflags = 0;
2718 uint8_t tos = 0;
2719 int ret = 0, ttl = 0;
2720 unsigned encap_limit = 0;
2721
2722 nlm = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_REPLACE | NLM_F_CREATE);
2723 if (!nlm)
2724 return -1;
2725
2726 nlmsg_append(nlm, &ifi, sizeof(ifi), 0);
2727 nla_put_string(nlm, IFLA_IFNAME, name);
2728
2729 struct nlattr *linkinfo = nla_nest_start(nlm, IFLA_LINKINFO);
2730 if (!linkinfo) {
2731 ret = -ENOMEM;
2732 goto failure;
2733 }
2734
2735 nla_put_string(nlm, IFLA_INFO_KIND, kind);
2736 struct nlattr *infodata = nla_nest_start(nlm, IFLA_INFO_DATA);
2737 if (!infodata) {
2738 ret = -ENOMEM;
2739 goto failure;
2740 }
2741
2742 if (link)
2743 nla_put_u32(nlm, IFLA_GRE_LINK, link);
2744
2745 if ((cur = tb[TUNNEL_ATTR_TTL]))
2746 ttl = blobmsg_get_u32(cur);
2747
2748 if ((cur = tb[TUNNEL_ATTR_TOS])) {
2749 char *str = blobmsg_get_string(cur);
2750 if (strcmp(str, "inherit")) {
2751 unsigned uval;
2752
2753 if (!system_tos_aton(str, &uval)) {
2754 ret = -EINVAL;
2755 goto failure;
2756 }
2757
2758 if (v6)
2759 flowinfo |= htonl(uval << 20) & IP6_FLOWINFO_TCLASS;
2760 else
2761 tos = uval;
2762 } else {
2763 if (v6)
2764 flags6 |= IP6_TNL_F_USE_ORIG_TCLASS;
2765 else
2766 tos = 1;
2767 }
2768 }
2769
2770 if ((cur = tb[TUNNEL_ATTR_DATA])) {
2771 struct blob_attr *tb_data[__GRE_DATA_ATTR_MAX];
2772
2773 blobmsg_parse(gre_data_attr_list.params, __GRE_DATA_ATTR_MAX, tb_data,
2774 blobmsg_data(cur), blobmsg_len(cur));
2775
2776 if ((cur = tb_data[GRE_DATA_IKEY])) {
2777 if ((ikey = blobmsg_get_u32(cur)))
2778 iflags |= GRE_KEY;
2779 }
2780
2781 if ((cur = tb_data[GRE_DATA_OKEY])) {
2782 if ((okey = blobmsg_get_u32(cur)))
2783 oflags |= GRE_KEY;
2784 }
2785
2786 if ((cur = tb_data[GRE_DATA_ICSUM])) {
2787 if (blobmsg_get_bool(cur))
2788 iflags |= GRE_CSUM;
2789 }
2790
2791 if ((cur = tb_data[GRE_DATA_OCSUM])) {
2792 if (blobmsg_get_bool(cur))
2793 oflags |= GRE_CSUM;
2794 }
2795
2796 if ((cur = tb_data[GRE_DATA_ISEQNO])) {
2797 if (blobmsg_get_bool(cur))
2798 iflags |= GRE_SEQ;
2799 }
2800
2801 if ((cur = tb_data[GRE_DATA_OSEQNO])) {
2802 if (blobmsg_get_bool(cur))
2803 oflags |= GRE_SEQ;
2804 }
2805
2806 if ((cur = tb_data[GRE_DATA_ENCAPLIMIT])) {
2807 char *str = blobmsg_get_string(cur);
2808
2809 if (strcmp(str, "ignore")) {
2810 char *e;
2811
2812 encap_limit = strtoul(str, &e, 0);
2813
2814 if (e == str || *e || encap_limit > 255) {
2815 ret = -EINVAL;
2816 goto failure;
2817 }
2818
2819 flags6 &= ~IP6_TNL_F_IGN_ENCAP_LIMIT;
2820 }
2821 }
2822 }
2823
2824 if (v6) {
2825 struct in6_addr in6buf;
2826 if ((cur = tb[TUNNEL_ATTR_LOCAL])) {
2827 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
2828 ret = -EINVAL;
2829 goto failure;
2830 }
2831 nla_put(nlm, IFLA_GRE_LOCAL, sizeof(in6buf), &in6buf);
2832 }
2833
2834 if ((cur = tb[TUNNEL_ATTR_REMOTE])) {
2835 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
2836 ret = -EINVAL;
2837 goto failure;
2838 }
2839 nla_put(nlm, IFLA_GRE_REMOTE, sizeof(in6buf), &in6buf);
2840 }
2841
2842 if (!(flags6 & IP6_TNL_F_IGN_ENCAP_LIMIT))
2843 nla_put_u8(nlm, IFLA_GRE_ENCAP_LIMIT, encap_limit);
2844
2845 if (flowinfo)
2846 nla_put_u32(nlm, IFLA_GRE_FLOWINFO, flowinfo);
2847
2848 if (flags6)
2849 nla_put_u32(nlm, IFLA_GRE_FLAGS, flags6);
2850
2851 if (!ttl)
2852 ttl = 64;
2853 } else {
2854 struct in_addr inbuf;
2855 bool set_df = true;
2856
2857 if ((cur = tb[TUNNEL_ATTR_LOCAL])) {
2858 if (inet_pton(AF_INET, blobmsg_data(cur), &inbuf) < 1) {
2859 ret = -EINVAL;
2860 goto failure;
2861 }
2862 nla_put(nlm, IFLA_GRE_LOCAL, sizeof(inbuf), &inbuf);
2863 }
2864
2865 if ((cur = tb[TUNNEL_ATTR_REMOTE])) {
2866 if (inet_pton(AF_INET, blobmsg_data(cur), &inbuf) < 1) {
2867 ret = -EINVAL;
2868 goto failure;
2869 }
2870 nla_put(nlm, IFLA_GRE_REMOTE, sizeof(inbuf), &inbuf);
2871
2872 if (IN_MULTICAST(ntohl(inbuf.s_addr))) {
2873 if (!okey) {
2874 okey = inbuf.s_addr;
2875 oflags |= GRE_KEY;
2876 }
2877
2878 if (!ikey) {
2879 ikey = inbuf.s_addr;
2880 iflags |= GRE_KEY;
2881 }
2882 }
2883 }
2884
2885 if ((cur = tb[TUNNEL_ATTR_DF]))
2886 set_df = blobmsg_get_bool(cur);
2887
2888 if (!set_df) {
2889 /* ttl != 0 and nopmtudisc are incompatible */
2890 if (ttl) {
2891 ret = -EINVAL;
2892 goto failure;
2893 }
2894 } else if (!ttl)
2895 ttl = 64;
2896
2897 nla_put_u8(nlm, IFLA_GRE_PMTUDISC, set_df ? 1 : 0);
2898
2899 nla_put_u8(nlm, IFLA_GRE_TOS, tos);
2900 }
2901
2902 if (ttl)
2903 nla_put_u8(nlm, IFLA_GRE_TTL, ttl);
2904
2905 if (oflags)
2906 nla_put_u16(nlm, IFLA_GRE_OFLAGS, oflags);
2907
2908 if (iflags)
2909 nla_put_u16(nlm, IFLA_GRE_IFLAGS, iflags);
2910
2911 if (okey)
2912 nla_put_u32(nlm, IFLA_GRE_OKEY, htonl(okey));
2913
2914 if (ikey)
2915 nla_put_u32(nlm, IFLA_GRE_IKEY, htonl(ikey));
2916
2917 nla_nest_end(nlm, infodata);
2918 nla_nest_end(nlm, linkinfo);
2919
2920 return system_rtnl_call(nlm);
2921
2922 failure:
2923 nlmsg_free(nlm);
2924 return ret;
2925 }
2926 #endif
2927
2928 #ifdef IFLA_VTI_MAX
2929 static int system_add_vti_tunnel(const char *name, const char *kind,
2930 const unsigned int link, struct blob_attr **tb, bool v6)
2931 {
2932 struct nl_msg *nlm;
2933 struct ifinfomsg ifi = { .ifi_family = AF_UNSPEC, };
2934 struct blob_attr *cur;
2935 int ret = 0;
2936
2937 nlm = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_REPLACE | NLM_F_CREATE);
2938 if (!nlm)
2939 return -1;
2940
2941 nlmsg_append(nlm, &ifi, sizeof(ifi), 0);
2942 nla_put_string(nlm, IFLA_IFNAME, name);
2943
2944 struct nlattr *linkinfo = nla_nest_start(nlm, IFLA_LINKINFO);
2945 if (!linkinfo) {
2946 ret = -ENOMEM;
2947 goto failure;
2948 }
2949
2950 nla_put_string(nlm, IFLA_INFO_KIND, kind);
2951 struct nlattr *infodata = nla_nest_start(nlm, IFLA_INFO_DATA);
2952 if (!infodata) {
2953 ret = -ENOMEM;
2954 goto failure;
2955 }
2956
2957 if (link)
2958 nla_put_u32(nlm, IFLA_VTI_LINK, link);
2959
2960 if (v6) {
2961 struct in6_addr in6buf;
2962 if ((cur = tb[TUNNEL_ATTR_LOCAL])) {
2963 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
2964 ret = -EINVAL;
2965 goto failure;
2966 }
2967 nla_put(nlm, IFLA_VTI_LOCAL, sizeof(in6buf), &in6buf);
2968 }
2969
2970 if ((cur = tb[TUNNEL_ATTR_REMOTE])) {
2971 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
2972 ret = -EINVAL;
2973 goto failure;
2974 }
2975 nla_put(nlm, IFLA_VTI_REMOTE, sizeof(in6buf), &in6buf);
2976 }
2977
2978 } else {
2979 struct in_addr inbuf;
2980
2981 if ((cur = tb[TUNNEL_ATTR_LOCAL])) {
2982 if (inet_pton(AF_INET, blobmsg_data(cur), &inbuf) < 1) {
2983 ret = -EINVAL;
2984 goto failure;
2985 }
2986 nla_put(nlm, IFLA_VTI_LOCAL, sizeof(inbuf), &inbuf);
2987 }
2988
2989 if ((cur = tb[TUNNEL_ATTR_REMOTE])) {
2990 if (inet_pton(AF_INET, blobmsg_data(cur), &inbuf) < 1) {
2991 ret = -EINVAL;
2992 goto failure;
2993 }
2994 nla_put(nlm, IFLA_VTI_REMOTE, sizeof(inbuf), &inbuf);
2995 }
2996
2997 }
2998
2999 if ((cur = tb[TUNNEL_ATTR_DATA])) {
3000 struct blob_attr *tb_data[__VTI_DATA_ATTR_MAX];
3001 uint32_t ikey = 0, okey = 0;
3002
3003 blobmsg_parse(vti_data_attr_list.params, __VTI_DATA_ATTR_MAX, tb_data,
3004 blobmsg_data(cur), blobmsg_len(cur));
3005
3006 if ((cur = tb_data[VTI_DATA_IKEY])) {
3007 if ((ikey = blobmsg_get_u32(cur)))
3008 nla_put_u32(nlm, IFLA_VTI_IKEY, htonl(ikey));
3009 }
3010
3011 if ((cur = tb_data[VTI_DATA_OKEY])) {
3012 if ((okey = blobmsg_get_u32(cur)))
3013 nla_put_u32(nlm, IFLA_VTI_OKEY, htonl(okey));
3014 }
3015 }
3016
3017 nla_nest_end(nlm, infodata);
3018 nla_nest_end(nlm, linkinfo);
3019
3020 return system_rtnl_call(nlm);
3021
3022 failure:
3023 nlmsg_free(nlm);
3024 return ret;
3025 }
3026 #endif
3027
3028 #ifdef IFLA_XFRM_MAX
3029 static int system_add_xfrm_tunnel(const char *name, const char *kind,
3030 const unsigned int link, struct blob_attr **tb)
3031 {
3032 struct nl_msg *nlm;
3033 struct ifinfomsg ifi = { .ifi_family = AF_UNSPEC, };
3034 struct blob_attr *cur;
3035 int ret = 0;
3036
3037 nlm = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_REPLACE | NLM_F_CREATE);
3038 if (!nlm)
3039 return -1;
3040
3041 nlmsg_append(nlm, &ifi, sizeof(ifi), 0);
3042 nla_put_string(nlm, IFLA_IFNAME, name);
3043
3044 struct nlattr *linkinfo = nla_nest_start(nlm, IFLA_LINKINFO);
3045 if (!linkinfo) {
3046 ret = -ENOMEM;
3047 goto failure;
3048 }
3049
3050 nla_put_string(nlm, IFLA_INFO_KIND, kind);
3051 struct nlattr *infodata = nla_nest_start(nlm, IFLA_INFO_DATA);
3052 if (!infodata) {
3053 ret = -ENOMEM;
3054 goto failure;
3055 }
3056
3057 if (link)
3058 nla_put_u32(nlm, IFLA_XFRM_LINK, link);
3059
3060 if ((cur = tb[TUNNEL_ATTR_DATA])) {
3061 struct blob_attr *tb_data[__XFRM_DATA_ATTR_MAX];
3062 uint32_t if_id = 0;
3063
3064 blobmsg_parse(xfrm_data_attr_list.params, __XFRM_DATA_ATTR_MAX, tb_data,
3065 blobmsg_data(cur), blobmsg_len(cur));
3066
3067 if ((cur = tb_data[XFRM_DATA_IF_ID])) {
3068 if ((if_id = blobmsg_get_u32(cur)))
3069 nla_put_u32(nlm, IFLA_XFRM_IF_ID, if_id);
3070 }
3071
3072 }
3073
3074 nla_nest_end(nlm, infodata);
3075 nla_nest_end(nlm, linkinfo);
3076
3077 return system_rtnl_call(nlm);
3078
3079 failure:
3080 nlmsg_free(nlm);
3081 return ret;
3082 }
3083 #endif
3084
3085 #ifdef IFLA_VXLAN_MAX
3086 static void system_vxlan_map_bool_attr(struct nl_msg *msg, struct blob_attr **tb_data, int attrtype, int vxlandatatype, bool invert) {
3087 struct blob_attr *cur;
3088 if ((cur = tb_data[vxlandatatype])) {
3089 bool val = blobmsg_get_bool(cur);
3090 if (invert)
3091 val = !val;
3092
3093 if ((attrtype == IFLA_VXLAN_GBP) && val)
3094 nla_put_flag(msg, attrtype);
3095 else
3096 nla_put_u8(msg, attrtype, val);
3097
3098 }
3099 }
3100
3101 static int system_add_vxlan(const char *name, const unsigned int link, struct blob_attr **tb, bool v6)
3102 {
3103 struct blob_attr *tb_data[__VXLAN_DATA_ATTR_MAX];
3104 struct nl_msg *msg;
3105 struct nlattr *linkinfo, *data;
3106 struct ifinfomsg iim = { .ifi_family = AF_UNSPEC, };
3107 struct blob_attr *cur;
3108 int ret = 0;
3109
3110 if ((cur = tb[TUNNEL_ATTR_DATA]))
3111 blobmsg_parse(vxlan_data_attr_list.params, __VXLAN_DATA_ATTR_MAX, tb_data,
3112 blobmsg_data(cur), blobmsg_len(cur));
3113 else
3114 return -EINVAL;
3115
3116 msg = nlmsg_alloc_simple(RTM_NEWLINK, NLM_F_REQUEST | NLM_F_CREATE | NLM_F_EXCL);
3117
3118 if (!msg)
3119 return -1;
3120
3121 nlmsg_append(msg, &iim, sizeof(iim), 0);
3122
3123 nla_put_string(msg, IFLA_IFNAME, name);
3124
3125 if ((cur = tb_data[VXLAN_DATA_ATTR_MACADDR])) {
3126 struct ether_addr *ea = ether_aton(blobmsg_get_string(cur));
3127 if (!ea) {
3128 ret = -EINVAL;
3129 goto failure;
3130 }
3131
3132 nla_put(msg, IFLA_ADDRESS, ETH_ALEN, ea);
3133 }
3134
3135 if ((cur = tb[TUNNEL_ATTR_MTU])) {
3136 uint32_t mtu = blobmsg_get_u32(cur);
3137 nla_put_u32(msg, IFLA_MTU, mtu);
3138 }
3139
3140 if (!(linkinfo = nla_nest_start(msg, IFLA_LINKINFO))) {
3141 ret = -ENOMEM;
3142 goto failure;
3143 }
3144
3145 nla_put_string(msg, IFLA_INFO_KIND, "vxlan");
3146
3147 if (!(data = nla_nest_start(msg, IFLA_INFO_DATA))) {
3148 ret = -ENOMEM;
3149 goto failure;
3150 }
3151
3152 if (link)
3153 nla_put_u32(msg, IFLA_VXLAN_LINK, link);
3154
3155 if ((cur = tb_data[VXLAN_DATA_ATTR_ID])) {
3156 uint32_t id = blobmsg_get_u32(cur);
3157 if (id >= (1u << 24) - 1) {
3158 ret = -EINVAL;
3159 goto failure;
3160 }
3161
3162 nla_put_u32(msg, IFLA_VXLAN_ID, id);
3163 }
3164
3165 if (v6) {
3166 struct in6_addr in6buf;
3167 if ((cur = tb[TUNNEL_ATTR_LOCAL])) {
3168 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
3169 ret = -EINVAL;
3170 goto failure;
3171 }
3172 nla_put(msg, IFLA_VXLAN_LOCAL6, sizeof(in6buf), &in6buf);
3173 }
3174
3175 if ((cur = tb[TUNNEL_ATTR_REMOTE])) {
3176 if (inet_pton(AF_INET6, blobmsg_data(cur), &in6buf) < 1) {
3177 ret = -EINVAL;
3178 goto failure;
3179 }
3180 nla_put(msg, IFLA_VXLAN_GROUP6, sizeof(in6buf), &in6buf);
3181 }
3182 } else {
3183 struct in_addr inbuf;
3184
3185 if ((cur = tb[TUNNEL_ATTR_LOCAL])) {
3186 if (inet_pton(AF_INET, blobmsg_data(cur), &inbuf) < 1) {
3187 ret = -EINVAL;
3188 goto failure;
3189 }
3190 nla_put(msg, IFLA_VXLAN_LOCAL, sizeof(inbuf), &inbuf);
3191 }
3192
3193 if ((cur = tb[TUNNEL_ATTR_REMOTE])) {
3194 if (inet_pton(AF_INET, blobmsg_data(cur), &inbuf) < 1) {
3195 ret = -EINVAL;
3196 goto failure;
3197 }
3198 nla_put(msg, IFLA_VXLAN_GROUP, sizeof(inbuf), &inbuf);
3199 }
3200 }
3201
3202 uint32_t port = 4789;
3203 if ((cur = tb_data[VXLAN_DATA_ATTR_PORT])) {
3204 port = blobmsg_get_u32(cur);
3205 if (port < 1 || port > 65535) {
3206 ret = -EINVAL;
3207 goto failure;
3208 }
3209 }
3210 nla_put_u16(msg, IFLA_VXLAN_PORT, htons(port));
3211
3212 if ((cur = tb_data[VXLAN_DATA_ATTR_SRCPORTMIN])) {
3213 struct ifla_vxlan_port_range srcports = {0,0};
3214
3215 uint32_t low = blobmsg_get_u32(cur);
3216 if (low < 1 || low > 65535 - 1) {
3217 ret = -EINVAL;
3218 goto failure;
3219 }
3220
3221 srcports.low = htons((uint16_t) low);
3222 srcports.high = htons((uint16_t) (low+1));
3223
3224 if ((cur = tb_data[VXLAN_DATA_ATTR_SRCPORTMAX])) {
3225 uint32_t high = blobmsg_get_u32(cur);
3226 if (high < 1 || high > 65535) {
3227 ret = -EINVAL;
3228 goto failure;
3229 }
3230
3231 if (high > low)
3232 srcports.high = htons((uint16_t) high);
3233 }
3234
3235 nla_put(msg, IFLA_VXLAN_PORT_RANGE, sizeof(srcports), &srcports);
3236 }
3237
3238 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_UDP_CSUM, VXLAN_DATA_ATTR_TXCSUM, false);
3239 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_UDP_ZERO_CSUM6_RX, VXLAN_DATA_ATTR_RXCSUM, true);
3240 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_UDP_ZERO_CSUM6_TX, VXLAN_DATA_ATTR_TXCSUM, true);
3241 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_LEARNING, VXLAN_DATA_ATTR_LEARNING, false);
3242 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_RSC , VXLAN_DATA_ATTR_RSC, false);
3243 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_PROXY , VXLAN_DATA_ATTR_PROXY, false);
3244 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_L2MISS , VXLAN_DATA_ATTR_L2MISS, false);
3245 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_L3MISS , VXLAN_DATA_ATTR_L3MISS, false);
3246 system_vxlan_map_bool_attr(msg, tb_data, IFLA_VXLAN_GBP , VXLAN_DATA_ATTR_GBP, false);
3247
3248 if ((cur = tb_data[VXLAN_DATA_ATTR_AGEING])) {
3249 uint32_t ageing = blobmsg_get_u32(cur);
3250 nla_put_u32(msg, IFLA_VXLAN_AGEING, ageing);
3251 }
3252
3253 if ((cur = tb_data[VXLAN_DATA_ATTR_LIMIT])) {
3254 uint32_t maxaddress = blobmsg_get_u32(cur);
3255 nla_put_u32(msg, IFLA_VXLAN_LIMIT, maxaddress);
3256 }
3257
3258 if ((cur = tb[TUNNEL_ATTR_TOS])) {
3259 char *str = blobmsg_get_string(cur);
3260 unsigned tos = 1;
3261
3262 if (strcmp(str, "inherit")) {
3263 if (!system_tos_aton(str, &tos)) {
3264 ret = -EINVAL;
3265 goto failure;
3266 }
3267 }
3268
3269 nla_put_u8(msg, IFLA_VXLAN_TOS, tos);
3270 }
3271
3272 if ((cur = tb[TUNNEL_ATTR_TTL])) {
3273 uint32_t ttl = blobmsg_get_u32(cur);
3274 if (ttl < 1 || ttl > 255) {
3275 ret = -EINVAL;
3276 goto failure;
3277 }
3278
3279 nla_put_u8(msg, IFLA_VXLAN_TTL, ttl);
3280 }
3281
3282 nla_nest_end(msg, data);
3283 nla_nest_end(msg, linkinfo);
3284
3285 ret = system_rtnl_call(msg);
3286 if (ret)
3287 D(SYSTEM, "Error adding vxlan '%s': %d\n", name, ret);
3288
3289 return ret;
3290
3291 failure:
3292 nlmsg_free(msg);
3293 return ret;
3294 }
3295 #endif
3296
3297 static int system_add_sit_tunnel(const char *name, const unsigned int link, struct blob_attr **tb)
3298 {
3299 struct blob_attr *cur;
3300 int ret = 0;
3301
3302 if (system_add_proto_tunnel(name, IPPROTO_IPV6, link, tb) < 0)
3303 return -1;
3304
3305 #ifdef SIOCADD6RD
3306 if ((cur = tb[TUNNEL_ATTR_DATA])) {
3307 struct blob_attr *tb_data[__SIXRD_DATA_ATTR_MAX];
3308 unsigned int mask;
3309 struct ip_tunnel_6rd p6;
3310
3311 blobmsg_parse(sixrd_data_attr_list.params, __SIXRD_DATA_ATTR_MAX, tb_data,
3312 blobmsg_data(cur), blobmsg_len(cur));
3313
3314 memset(&p6, 0, sizeof(p6));
3315
3316 if ((cur = tb_data[SIXRD_DATA_PREFIX])) {
3317 if (!parse_ip_and_netmask(AF_INET6, blobmsg_data(cur),
3318 &p6.prefix, &mask) || mask > 128) {
3319 ret = -EINVAL;
3320 goto failure;
3321 }
3322
3323 p6.prefixlen = mask;
3324 }
3325
3326 if ((cur = tb_data[SIXRD_DATA_RELAY_PREFIX])) {
3327 if (!parse_ip_and_netmask(AF_INET, blobmsg_data(cur),
3328 &p6.relay_prefix, &mask) || mask > 32) {
3329 ret = -EINVAL;
3330 goto failure;
3331 }
3332
3333 p6.relay_prefixlen = mask;
3334 }
3335
3336 if (tunnel_ioctl(name, SIOCADD6RD, &p6) < 0) {
3337 ret = -1;
3338 goto failure;
3339 }
3340 }
3341 #endif
3342
3343 return ret;
3344
3345 failure:
3346 __system_del_ip_tunnel(name, tb);
3347 return ret;
3348 }
3349
3350 static int system_add_proto_tunnel(const char *name, const uint8_t proto, const unsigned int link, struct blob_attr **tb)
3351 {
3352 struct blob_attr *cur;
3353 bool set_df = true;
3354 struct ip_tunnel_parm p = {
3355 .link = link,
3356 .iph = {
3357 .version = 4,
3358 .ihl = 5,
3359 .protocol = proto,
3360 }
3361 };
3362
3363 if ((cur = tb[TUNNEL_ATTR_LOCAL]) &&
3364 inet_pton(AF_INET, blobmsg_data(cur), &p.iph.saddr) < 1)
3365 return -EINVAL;
3366
3367 if ((cur = tb[TUNNEL_ATTR_REMOTE]) &&
3368 inet_pton(AF_INET, blobmsg_data(cur), &p.iph.daddr) < 1)
3369 return -EINVAL;
3370
3371 if ((cur = tb[TUNNEL_ATTR_DF]))
3372 set_df = blobmsg_get_bool(cur);
3373
3374 if ((cur = tb[TUNNEL_ATTR_TTL]))
3375 p.iph.ttl = blobmsg_get_u32(cur);
3376
3377 if ((cur = tb[TUNNEL_ATTR_TOS])) {
3378 char *str = blobmsg_get_string(cur);
3379 if (strcmp(str, "inherit")) {
3380 unsigned uval;
3381
3382 if (!system_tos_aton(str, &uval))
3383 return -EINVAL;
3384
3385 p.iph.tos = uval;
3386 } else
3387 p.iph.tos = 1;
3388 }
3389
3390 p.iph.frag_off = set_df ? htons(IP_DF) : 0;
3391 /* ttl !=0 and nopmtudisc are incompatible */
3392 if (p.iph.ttl && p.iph.frag_off == 0)
3393 return -EINVAL;
3394
3395 strncpy(p.name, name, sizeof(p.name) - 1);
3396
3397 switch (p.iph.protocol) {
3398 case IPPROTO_IPIP:
3399 return tunnel_ioctl("tunl0", SIOCADDTUNNEL, &p);
3400 case IPPROTO_IPV6:
3401 return tunnel_ioctl("sit0", SIOCADDTUNNEL, &p);
3402 default:
3403 break;
3404 }
3405 return -1;
3406 }
3407
3408 static int __system_del_ip_tunnel(const char *name, struct blob_attr **tb)
3409 {
3410 struct blob_attr *cur;
3411 const char *str;
3412
3413 if (!(cur = tb[TUNNEL_ATTR_TYPE]))
3414 return -EINVAL;
3415 str = blobmsg_data(cur);
3416
3417 if (!strcmp(str, "greip") || !strcmp(str, "gretapip") ||
3418 !strcmp(str, "greip6") || !strcmp(str, "gretapip6") ||
3419 !strcmp(str, "vtiip") || !strcmp(str, "vtiip6") ||
3420 !strcmp(str, "vxlan") || !strcmp(str, "vxlan6") ||
3421 !strcmp(str, "xfrm"))
3422 return system_link_del(name);
3423 else
3424 return tunnel_ioctl(name, SIOCDELTUNNEL, NULL);
3425 }
3426
3427 int system_del_ip_tunnel(const char *name, struct blob_attr *attr)
3428 {
3429 struct blob_attr *tb[__TUNNEL_ATTR_MAX];
3430
3431 blobmsg_parse(tunnel_attr_list.params, __TUNNEL_ATTR_MAX, tb,
3432 blob_data(attr), blob_len(attr));
3433
3434 return __system_del_ip_tunnel(name, tb);
3435 }
3436
3437 int system_update_ipv6_mtu(struct device *dev, int mtu)
3438 {
3439 int ret = -1;
3440 char buf[64];
3441 int fd;
3442
3443 snprintf(buf, sizeof(buf), "/proc/sys/net/ipv6/conf/%s/mtu",
3444 dev->ifname);
3445
3446 fd = open(buf, O_RDWR);
3447 if (fd < 0)
3448 return ret;
3449
3450 if (!mtu) {
3451 ssize_t len = read(fd, buf, sizeof(buf) - 1);
3452 if (len < 0)
3453 goto out;
3454
3455 buf[len] = 0;
3456 ret = atoi(buf);
3457 } else {
3458 if (write(fd, buf, snprintf(buf, sizeof(buf), "%i", mtu)) > 0)
3459 ret = mtu;
3460 }
3461
3462 out:
3463 close(fd);
3464 return ret;
3465 }
3466
3467 int system_add_ip_tunnel(const char *name, struct blob_attr *attr)
3468 {
3469 struct blob_attr *tb[__TUNNEL_ATTR_MAX];
3470 struct blob_attr *cur;
3471 const char *str;
3472
3473 blobmsg_parse(tunnel_attr_list.params, __TUNNEL_ATTR_MAX, tb,
3474 blob_data(attr), blob_len(attr));
3475
3476 __system_del_ip_tunnel(name, tb);
3477
3478 if (!(cur = tb[TUNNEL_ATTR_TYPE]))
3479 return -EINVAL;
3480 str = blobmsg_data(cur);
3481
3482 unsigned int ttl = 0;
3483 if ((cur = tb[TUNNEL_ATTR_TTL])) {
3484 ttl = blobmsg_get_u32(cur);
3485 if (ttl > 255)
3486 return -EINVAL;
3487 }
3488
3489 unsigned int link = 0;
3490 if ((cur = tb[TUNNEL_ATTR_LINK])) {
3491 struct interface *iface = vlist_find(&interfaces, blobmsg_data(cur), iface, node);
3492 if (!iface)
3493 return -EINVAL;
3494
3495 if (iface->l3_dev.dev)
3496 link = iface->l3_dev.dev->ifindex;
3497 }
3498
3499 if (!strcmp(str, "sit"))
3500 return system_add_sit_tunnel(name, link, tb);
3501 #ifdef IFLA_IPTUN_MAX
3502 else if (!strcmp(str, "ipip6")) {
3503 return system_add_ip6_tunnel(name, link, tb);
3504 } else if (!strcmp(str, "greip")) {
3505 return system_add_gre_tunnel(name, "gre", link, tb, false);
3506 } else if (!strcmp(str, "gretapip")) {
3507 return system_add_gre_tunnel(name, "gretap", link, tb, false);
3508 } else if (!strcmp(str, "greip6")) {
3509 return system_add_gre_tunnel(name, "ip6gre", link, tb, true);
3510 } else if (!strcmp(str, "gretapip6")) {
3511 return system_add_gre_tunnel(name, "ip6gretap", link, tb, true);
3512 #ifdef IFLA_VTI_MAX
3513 } else if (!strcmp(str, "vtiip")) {
3514 return system_add_vti_tunnel(name, "vti", link, tb, false);
3515 } else if (!strcmp(str, "vtiip6")) {
3516 return system_add_vti_tunnel(name, "vti6", link, tb, true);
3517 #endif
3518 #ifdef IFLA_XFRM_MAX
3519 } else if (!strcmp(str, "xfrm")) {
3520 return system_add_xfrm_tunnel(name, "xfrm", link, tb);
3521 #endif
3522 #ifdef IFLA_VXLAN_MAX
3523 } else if(!strcmp(str, "vxlan")) {
3524 return system_add_vxlan(name, link, tb, false);
3525 } else if(!strcmp(str, "vxlan6")) {
3526 return system_add_vxlan(name, link, tb, true);
3527 #endif
3528 #endif
3529 } else if (!strcmp(str, "ipip")) {
3530 return system_add_proto_tunnel(name, IPPROTO_IPIP, link, tb);
3531 }
3532 else
3533 return -EINVAL;
3534
3535 return 0;
3536 }
OpenWrt Network interface configuration daemon