2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2022 Alexander V. Chernikov <melifaro@FreeBSD.org>
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
15 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28 #include <sys/cdefs.h>
29 __FBSDID("$FreeBSD$");
31 #include "opt_inet6.h"
32 #include <sys/types.h>
33 #include <sys/eventhandler.h>
34 #include <sys/kernel.h>
35 #include <sys/malloc.h>
36 #include <sys/socket.h>
37 #include <sys/sockio.h>
38 #include <sys/syslog.h>
41 #include <net/if_dl.h>
42 #include <net/if_media.h>
43 #include <net/if_var.h>
44 #include <net/if_clone.h>
45 #include <net/route.h>
46 #include <net/route/nhop.h>
47 #include <net/route/route_ctl.h>
48 #include <netlink/netlink.h>
49 #include <netlink/netlink_ctl.h>
50 #include <netlink/netlink_route.h>
51 #include <netlink/route/route_var.h>
53 #include <netinet6/scope6_var.h> /* scope deembedding */
55 #define DEBUG_MOD_NAME nl_iface
56 #define DEBUG_MAX_LEVEL LOG_DEBUG3
57 #include <netlink/netlink_debug.h>
58 _DECLARE_DEBUG(LOG_DEBUG);
60 struct netlink_walkargs {
71 static eventhandler_tag ifdetach_event, ifattach_event, iflink_event, ifaddr_event;
73 static SLIST_HEAD(, nl_cloner) nl_cloners = SLIST_HEAD_INITIALIZER(nl_cloners);
75 static struct sx rtnl_cloner_lock;
76 SX_SYSINIT(rtnl_cloner_lock, &rtnl_cloner_lock, "rtnl cloner lock");
78 static struct nl_cloner *rtnl_iface_find_cloner_locked(const char *name);
82 * sendto(3, {{len=32, type=RTM_GETLINK, flags=NLM_F_REQUEST|NLM_F_DUMP, seq=1641940952, pid=0},
83 * {ifi_family=AF_INET, ifi_type=ARPHRD_NETROM, ifi_index=0, ifi_flags=0, ifi_change=0}}, 32, 0, NULL, 0) = 32
86 * {ifi_family=AF_UNSPEC, ifi_type=ARPHRD_ETHER, ifi_index=if_nametoindex("enp0s31f6"), ifi_flags=IFF_UP|IFF_BROADCAST|IFF_RUNNING|IFF_MULTICAST|IFF_LOWER_UP, ifi_change=0},
87 {{nla_len=10, nla_type=IFLA_ADDRESS}, "\xfe\x54\x00\x52\x3e\x90"}
90 {{nla_len=14, nla_type=IFLA_IFNAME}, "enp0s31f6"},
91 {{nla_len=8, nla_type=IFLA_TXQLEN}, 1000},
92 {{nla_len=5, nla_type=IFLA_OPERSTATE}, 6},
93 {{nla_len=5, nla_type=IFLA_LINKMODE}, 0},
94 {{nla_len=8, nla_type=IFLA_MTU}, 1500},
95 {{nla_len=8, nla_type=IFLA_MIN_MTU}, 68},
96 {{nla_len=8, nla_type=IFLA_MAX_MTU}, 9000},
97 {{nla_len=8, nla_type=IFLA_GROUP}, 0},
98 {{nla_len=8, nla_type=IFLA_PROMISCUITY}, 0},
99 {{nla_len=8, nla_type=IFLA_NUM_TX_QUEUES}, 1},
100 {{nla_len=8, nla_type=IFLA_GSO_MAX_SEGS}, 65535},
101 {{nla_len=8, nla_type=IFLA_GSO_MAX_SIZE}, 65536},
102 {{nla_len=8, nla_type=IFLA_NUM_RX_QUEUES}, 1},
103 {{nla_len=5, nla_type=IFLA_CARRIER}, 1},
104 {{nla_len=13, nla_type=IFLA_QDISC}, "fq_codel"},
105 {{nla_len=8, nla_type=IFLA_CARRIER_CHANGES}, 2},
106 {{nla_len=5, nla_type=IFLA_PROTO_DOWN}, 0},
107 {{nla_len=8, nla_type=IFLA_CARRIER_UP_COUNT}, 1},
108 {{nla_len=8, nla_type=IFLA_CARRIER_DOWN_COUNT}, 1},
112 uint8_t ifla_operstate;
113 uint8_t ifla_carrier;
117 get_operstate_ether(struct ifnet *ifp, struct if_state *pstate)
119 struct ifmediareq ifmr = {};
121 error = (*ifp->if_ioctl)(ifp, SIOCGIFMEDIA, (void *)&ifmr);
124 NL_LOG(LOG_DEBUG, "error calling SIOCGIFMEDIA on %s: %d",
125 if_name(ifp), error);
129 switch (IFM_TYPE(ifmr.ifm_active)) {
131 if (ifmr.ifm_status & IFM_ACTIVE) {
132 pstate->ifla_carrier = 1;
133 if (ifp->if_flags & IFF_MONITOR)
134 pstate->ifla_operstate = IF_OPER_DORMANT;
136 pstate->ifla_operstate = IF_OPER_UP;
138 pstate->ifla_operstate = IF_OPER_DOWN;
143 get_stats(struct nl_writer *nw, struct ifnet *ifp)
145 struct rtnl_link_stats64 *stats;
147 int nla_len = sizeof(struct nlattr) + sizeof(*stats);
148 struct nlattr *nla = nlmsg_reserve_data(nw, nla_len, struct nlattr);
151 nla->nla_type = IFLA_STATS64;
152 nla->nla_len = nla_len;
153 stats = (struct rtnl_link_stats64 *)(nla + 1);
155 stats->rx_packets = ifp->if_get_counter(ifp, IFCOUNTER_IPACKETS);
156 stats->tx_packets = ifp->if_get_counter(ifp, IFCOUNTER_OPACKETS);
157 stats->rx_bytes = ifp->if_get_counter(ifp, IFCOUNTER_IBYTES);
158 stats->tx_bytes = ifp->if_get_counter(ifp, IFCOUNTER_OBYTES);
159 stats->rx_errors = ifp->if_get_counter(ifp, IFCOUNTER_IERRORS);
160 stats->tx_errors = ifp->if_get_counter(ifp, IFCOUNTER_OERRORS);
161 stats->rx_dropped = ifp->if_get_counter(ifp, IFCOUNTER_IQDROPS);
162 stats->tx_dropped = ifp->if_get_counter(ifp, IFCOUNTER_OQDROPS);
163 stats->multicast = ifp->if_get_counter(ifp, IFCOUNTER_IMCASTS);
164 stats->rx_nohandler = ifp->if_get_counter(ifp, IFCOUNTER_NOPROTO);
170 get_operstate(struct ifnet *ifp, struct if_state *pstate)
172 pstate->ifla_operstate = IF_OPER_UNKNOWN;
173 pstate->ifla_carrier = 0; /* no carrier */
175 switch (ifp->if_type) {
178 get_operstate_ether(ifp, pstate);
181 /* Map admin state to the operstate */
182 if (ifp->if_flags & IFF_UP) {
183 pstate->ifla_operstate = IF_OPER_UP;
184 pstate->ifla_carrier = 1;
186 pstate->ifla_operstate = IF_OPER_DOWN;
192 ifp_flags_to_netlink(const struct ifnet *ifp)
194 return (ifp->if_flags | ifp->if_drv_flags);
197 #define LLADDR_CONST(s) ((const void *)((s)->sdl_data + (s)->sdl_nlen))
199 dump_sa(struct nl_writer *nw, int attr, const struct sockaddr *sa)
201 uint32_t addr_len = 0;
202 const void *addr_data = NULL;
204 struct in6_addr addr6;
210 switch (sa->sa_family) {
213 addr_len = sizeof(struct in_addr);
214 addr_data = &((const struct sockaddr_in *)sa)->sin_addr;
219 in6_splitscope(&((const struct sockaddr_in6 *)sa)->sin6_addr, &addr6, &addr_len);
220 addr_len = sizeof(struct in6_addr);
225 addr_len = ((const struct sockaddr_dl *)sa)->sdl_alen;
226 addr_data = LLADDR_CONST((const struct sockaddr_dl *)sa);
229 NL_LOG(LOG_DEBUG, "unsupported family: %d, skipping", sa->sa_family);
233 return (nlattr_add(nw, attr, addr_len, addr_data));
237 * Dumps interface state, properties and metrics.
238 * @nw: message writer
239 * @ifp: target interface
240 * @hdr: template header
241 * @if_flags_mask: changed if_[drv]_flags bitmask
243 * This function is called without epoch and MAY sleep.
246 dump_iface(struct nl_writer *nw, struct ifnet *ifp, const struct nlmsghdr *hdr,
249 struct ifinfomsg *ifinfo;
251 NL_LOG(LOG_DEBUG3, "dumping interface %s data", if_name(ifp));
253 if (!nlmsg_reply(nw, hdr, sizeof(struct ifinfomsg)))
256 ifinfo = nlmsg_reserve_object(nw, struct ifinfomsg);
257 ifinfo->ifi_family = AF_UNSPEC;
258 ifinfo->__ifi_pad = 0;
259 ifinfo->ifi_type = ifp->if_type;
260 ifinfo->ifi_index = ifp->if_index;
261 ifinfo->ifi_flags = ifp_flags_to_netlink(ifp);
262 ifinfo->ifi_change = if_flags_mask;
264 struct if_state ifs = {};
265 get_operstate(ifp, &ifs);
267 if (ifs.ifla_operstate == IF_OPER_UP)
268 ifinfo->ifi_flags |= IFF_LOWER_UP;
270 nlattr_add_string(nw, IFLA_IFNAME, if_name(ifp));
271 nlattr_add_u8(nw, IFLA_OPERSTATE, ifs.ifla_operstate);
272 nlattr_add_u8(nw, IFLA_CARRIER, ifs.ifla_carrier);
275 nlattr_add_u8(nw, IFLA_PROTO_DOWN, val);
276 nlattr_add_u8(nw, IFLA_LINKMODE, val);
278 if ((ifp->if_addr != NULL)) {
279 dump_sa(nw, IFLA_ADDRESS, ifp->if_addr->ifa_addr);
282 if ((ifp->if_broadcastaddr != NULL)) {
283 nlattr_add(nw, IFLA_BROADCAST, ifp->if_addrlen,
284 ifp->if_broadcastaddr);
287 nlattr_add_u32(nw, IFLA_MTU, ifp->if_mtu);
289 nlattr_add_u32(nw, IFLA_MIN_MTU, 60);
290 nlattr_add_u32(nw, IFLA_MAX_MTU, 9000);
291 nlattr_add_u32(nw, IFLA_GROUP, 0);
294 if (ifp->if_description != NULL)
295 nlattr_add_string(nw, IFLA_IFALIAS, ifp->if_description);
299 uint32_t val = (ifp->if_flags & IFF_PROMISC) != 0;
300 nlattr_add_u32(nw, IFLA_PROMISCUITY, val);
302 sx_slock(&rtnl_cloner_lock);
303 struct nl_cloner *cloner = rtnl_iface_find_cloner_locked(ifp->if_dname);
304 if (cloner != NULL && cloner->dump_f != NULL) {
305 /* Ignore any dump error */
306 cloner->dump_f(ifp, nw);
308 sx_sunlock(&rtnl_cloner_lock);
314 NL_LOG(LOG_DEBUG, "unable to dump interface %s state (ENOMEM)", if_name(ifp));
320 check_ifmsg(void *hdr, struct nl_pstate *npt)
322 struct ifinfomsg *ifm = hdr;
324 if (ifm->__ifi_pad != 0 || ifm->ifi_type != 0 ||
325 ifm->ifi_flags != 0 || ifm->ifi_change != 0) {
326 nlmsg_report_err_msg(npt,
327 "strict checking: non-zero values in ifinfomsg header");
334 #define _IN(_field) offsetof(struct ifinfomsg, _field)
335 #define _OUT(_field) offsetof(struct nl_parsed_link, _field)
336 static const struct nlfield_parser nlf_p_if[] = {
337 { .off_in = _IN(ifi_type), .off_out = _OUT(ifi_type), .cb = nlf_get_u16 },
338 { .off_in = _IN(ifi_index), .off_out = _OUT(ifi_index), .cb = nlf_get_u32 },
339 { .off_in = _IN(ifi_flags), .off_out = _OUT(ifi_flags), .cb = nlf_get_u32 },
340 { .off_in = _IN(ifi_change), .off_out = _OUT(ifi_change), .cb = nlf_get_u32 },
343 static const struct nlattr_parser nla_p_linfo[] = {
344 { .type = IFLA_INFO_KIND, .off = _OUT(ifla_cloner), .cb = nlattr_get_stringn },
345 { .type = IFLA_INFO_DATA, .off = _OUT(ifla_idata), .cb = nlattr_get_nla },
347 NL_DECLARE_ATTR_PARSER(linfo_parser, nla_p_linfo);
349 static const struct nlattr_parser nla_p_if[] = {
350 { .type = IFLA_IFNAME, .off = _OUT(ifla_ifname), .cb = nlattr_get_string },
351 { .type = IFLA_MTU, .off = _OUT(ifla_mtu), .cb = nlattr_get_uint32 },
352 { .type = IFLA_LINK, .off = _OUT(ifi_index), .cb = nlattr_get_uint32 },
353 { .type = IFLA_LINKINFO, .arg = &linfo_parser, .cb = nlattr_get_nested },
354 { .type = IFLA_IFALIAS, .off = _OUT(ifla_ifalias), .cb = nlattr_get_string },
355 { .type = IFLA_GROUP, .off = _OUT(ifla_group), .cb = nlattr_get_string },
356 { .type = IFLA_ALT_IFNAME, .off = _OUT(ifla_ifname), .cb = nlattr_get_string },
360 NL_DECLARE_STRICT_PARSER(ifmsg_parser, struct ifinfomsg, check_ifmsg, nlf_p_if, nla_p_if);
363 match_iface(struct nl_parsed_link *attrs, struct ifnet *ifp)
365 if (attrs->ifi_index != 0 && attrs->ifi_index != ifp->if_index)
367 if (attrs->ifi_type != 0 && attrs->ifi_index != ifp->if_type)
369 if (attrs->ifla_ifname != NULL && strcmp(attrs->ifla_ifname, if_name(ifp)))
371 /* TODO: add group match */
377 * {nlmsg_len=52, nlmsg_type=RTM_GETLINK, nlmsg_flags=NLM_F_REQUEST, nlmsg_seq=1662842818, nlmsg_pid=0},
378 * {ifi_family=AF_PACKET, ifi_type=ARPHRD_NETROM, ifi_index=0, ifi_flags=0, ifi_change=0},
380 * [{nla_len=10, nla_type=IFLA_IFNAME}, "vnet9"],
381 * [{nla_len=8, nla_type=IFLA_EXT_MASK}, RTEXT_FILTER_VF]
385 rtnl_handle_getlink(struct nlmsghdr *hdr, struct nlpcb *nlp, struct nl_pstate *npt)
387 struct epoch_tracker et;
391 struct nl_parsed_link attrs = {};
392 error = nl_parse_nlmsg(hdr, &ifmsg_parser, npt, &attrs);
396 struct netlink_walkargs wa = {
399 .hdr.nlmsg_pid = hdr->nlmsg_pid,
400 .hdr.nlmsg_seq = hdr->nlmsg_seq,
401 .hdr.nlmsg_flags = hdr->nlmsg_flags,
402 .hdr.nlmsg_type = NL_RTM_NEWLINK,
405 /* Fast track for an interface w/ explicit name or index match */
406 if ((attrs.ifi_index != 0) || (attrs.ifla_ifname != NULL)) {
407 if (attrs.ifi_index != 0) {
408 NLP_LOG(LOG_DEBUG3, nlp, "fast track -> searching index %u",
411 ifp = ifnet_byindex_ref(attrs.ifi_index);
414 NLP_LOG(LOG_DEBUG3, nlp, "fast track -> searching name %s",
416 ifp = ifunit_ref(attrs.ifla_ifname);
420 if (match_iface(&attrs, ifp)) {
421 if (!dump_iface(wa.nw, ifp, &wa.hdr, 0))
431 /* Always treat non-direct-match as a multipart message */
432 wa.hdr.nlmsg_flags |= NLM_F_MULTI;
435 * Fetching some link properties require performing ioctl's that may be blocking.
436 * Address it by saving referenced pointers of the matching links,
437 * exiting from epoch and going through the list one-by-one.
440 NL_LOG(LOG_DEBUG2, "Start dump");
442 struct ifnet **match_array;
443 int offset = 0, base_count = 16; /* start with 128 bytes */
444 match_array = malloc(base_count * sizeof(void *), M_TEMP, M_NOWAIT);
446 NLP_LOG(LOG_DEBUG3, nlp, "MATCHING: index=%u type=%d name=%s",
447 attrs.ifi_index, attrs.ifi_type, attrs.ifla_ifname);
449 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
451 if (match_iface(&attrs, ifp)) {
452 if (offset >= base_count) {
453 /* Too many matches, need to reallocate */
454 struct ifnet **new_array;
455 int sz = base_count * sizeof(void *);
457 new_array = malloc(sz * 2, M_TEMP, M_NOWAIT);
458 if (new_array == NULL) {
462 memcpy(new_array, match_array, sz);
463 free(match_array, M_TEMP);
464 match_array = new_array;
468 match_array[offset++] = ifp;
473 NL_LOG(LOG_DEBUG2, "Matched %d interface(s), dumping", offset);
474 for (int i = 0; error == 0 && i < offset; i++) {
475 if (!dump_iface(wa.nw, match_array[i], &wa.hdr, 0))
478 for (int i = 0; i < offset; i++)
479 if_rele(match_array[i]);
480 free(match_array, M_TEMP);
482 NL_LOG(LOG_DEBUG2, "End dump, iterated %d dumped %d", wa.count, wa.dumped);
484 if (!nlmsg_end_dump(wa.nw, error, &wa.hdr)) {
485 NL_LOG(LOG_DEBUG, "Unable to finalize the dump");
493 * sendmsg(3, {msg_name={sa_family=AF_NETLINK, nl_pid=0, nl_groups=00000000}, msg_namelen=12, msg_iov=[{iov_base=[
494 * {nlmsg_len=60, nlmsg_type=RTM_NEWLINK, nlmsg_flags=NLM_F_REQUEST|NLM_F_ACK|NLM_F_EXCL|NLM_F_CREATE, nlmsg_seq=1662715618, nlmsg_pid=0},
495 * {ifi_family=AF_UNSPEC, ifi_type=ARPHRD_NETROM, ifi_index=0, ifi_flags=0, ifi_change=0},
496 * {nla_len=11, nla_type=IFLA_IFNAME}, "dummy0"],
498 * {nla_len=16, nla_type=IFLA_LINKINFO},
500 * {nla_len=9, nla_type=IFLA_INFO_KIND}, "dummy"...
506 rtnl_handle_dellink(struct nlmsghdr *hdr, struct nlpcb *nlp, struct nl_pstate *npt)
508 struct epoch_tracker et;
512 struct nl_parsed_link attrs = {};
513 error = nl_parse_nlmsg(hdr, &ifmsg_parser, npt, &attrs);
518 ifp = ifnet_byindex_ref(attrs.ifi_index);
521 NLP_LOG(LOG_DEBUG, nlp, "unable to find interface %u", attrs.ifi_index);
524 NLP_LOG(LOG_DEBUG3, nlp, "mapped ifindex %u to %s", attrs.ifi_index, if_name(ifp));
526 sx_xlock(&ifnet_detach_sxlock);
527 error = if_clone_destroy(if_name(ifp));
528 sx_xunlock(&ifnet_detach_sxlock);
530 NLP_LOG(LOG_DEBUG2, nlp, "deleting interface %s returned %d", if_name(ifp), error);
538 * type=RTM_NEWLINK, flags=NLM_F_REQUEST|NLM_F_ACK|NLM_F_EXCL|NLM_F_CREATE, seq=1668185590, pid=0},
539 * {ifi_family=AF_UNSPEC, ifi_type=ARPHRD_NETROM, ifi_index=0, ifi_flags=0, ifi_change=0}
541 * {{nla_len=8, nla_type=IFLA_MTU}, 123},
542 * {{nla_len=10, nla_type=IFLA_IFNAME}, "vlan1"},
543 * {{nla_len=24, nla_type=IFLA_LINKINFO},
545 * {{nla_len=8, nla_type=IFLA_INFO_KIND}, "vlan"...},
546 * {{nla_len=12, nla_type=IFLA_INFO_DATA}, "\x06\x00\x01\x00\x7b\x00\x00\x00"}]}]}
549 * type=RTM_NEWLINK, flags=NLM_F_REQUEST|NLM_F_ACK, seq=1668185923, pid=0},
550 * {ifi_family=AF_UNSPEC, ifi_type=ARPHRD_NETROM, ifi_index=if_nametoindex("lo"), ifi_flags=0, ifi_change=0},
551 * {{nla_len=8, nla_type=IFLA_MTU}, 123}}
554 * Check command availability:
555 * type=RTM_NEWLINK, flags=NLM_F_REQUEST|NLM_F_ACK, seq=0, pid=0},
556 * {ifi_family=AF_UNSPEC, ifi_type=ARPHRD_NETROM, ifi_index=0, ifi_flags=0, ifi_change=0}
561 create_link(struct nlmsghdr *hdr, struct nl_parsed_link *lattrs,
562 struct nlattr_bmask *bm, struct nlpcb *nlp, struct nl_pstate *npt)
564 if (lattrs->ifla_ifname == NULL || strlen(lattrs->ifla_ifname) == 0) {
565 NLMSG_REPORT_ERR_MSG(npt, "empty IFLA_IFNAME attribute");
568 if (lattrs->ifla_cloner == NULL || strlen(lattrs->ifla_cloner) == 0) {
569 NLMSG_REPORT_ERR_MSG(npt, "empty IFLA_INFO_KIND attribute");
576 sx_slock(&rtnl_cloner_lock);
577 struct nl_cloner *cloner = rtnl_iface_find_cloner_locked(lattrs->ifla_cloner);
578 if (cloner != NULL) {
580 error = cloner->create_f(lattrs, bm, nlp, npt);
582 sx_sunlock(&rtnl_cloner_lock);
585 error = generic_cloner.create_f(lattrs, bm, nlp, npt);
591 modify_link(struct nlmsghdr *hdr, struct nl_parsed_link *lattrs,
592 struct nlattr_bmask *bm, struct nlpcb *nlp, struct nl_pstate *npt)
594 struct ifnet *ifp = NULL;
595 struct epoch_tracker et;
597 if (lattrs->ifi_index == 0 && lattrs->ifla_ifname == NULL) {
599 * Applications like ip(8) verify RTM_NEWLINK command
600 * existence by calling it with empty arguments. Always
601 * return "innocent" error in that case.
603 NLMSG_REPORT_ERR_MSG(npt, "empty ifi_index field");
607 if (lattrs->ifi_index != 0) {
609 ifp = ifnet_byindex_ref(lattrs->ifi_index);
612 NLMSG_REPORT_ERR_MSG(npt, "unable to find interface #%u",
618 if (ifp == NULL && lattrs->ifla_ifname != NULL) {
619 ifp = ifunit_ref(lattrs->ifla_ifname);
621 NLMSG_REPORT_ERR_MSG(npt, "unable to find interface %s",
622 lattrs->ifla_ifname);
630 * There can be multiple kinds of interfaces:
631 * 1) cloned, with additional options
632 * 2) cloned, but w/o additional options
633 * 3) non-cloned (e.g. "physical).
635 * Thus, try to find cloner-specific callback and fallback to the
636 * "default" handler if not found.
641 sx_slock(&rtnl_cloner_lock);
642 struct nl_cloner *cloner = rtnl_iface_find_cloner_locked(ifp->if_dname);
643 if (cloner != NULL) {
645 error = cloner->modify_f(ifp, lattrs, bm, nlp, npt);
647 sx_sunlock(&rtnl_cloner_lock);
650 error = generic_cloner.modify_f(ifp, lattrs, bm, nlp, npt);
659 rtnl_handle_newlink(struct nlmsghdr *hdr, struct nlpcb *nlp, struct nl_pstate *npt)
661 struct nlattr_bmask bm;
664 struct nl_parsed_link attrs = {};
665 error = nl_parse_nlmsg(hdr, &ifmsg_parser, npt, &attrs);
668 nl_get_attrs_bmask_nlmsg(hdr, &ifmsg_parser, &bm);
670 if (hdr->nlmsg_flags & NLM_F_CREATE)
671 return (create_link(hdr, &attrs, &bm, nlp, npt));
673 return (modify_link(hdr, &attrs, &bm, nlp, npt));
676 struct nl_parsed_ifa {
678 uint8_t ifa_prefixlen;
682 struct sockaddr *ifa_address;
683 struct sockaddr *ifa_local;
686 #define _IN(_field) offsetof(struct ifaddrmsg, _field)
687 #define _OUT(_field) offsetof(struct nl_parsed_ifa, _field)
688 static const struct nlfield_parser nlf_p_ifa[] = {
689 { .off_in = _IN(ifa_family), .off_out = _OUT(ifa_family), .cb = nlf_get_u8 },
690 { .off_in = _IN(ifa_prefixlen), .off_out = _OUT(ifa_prefixlen), .cb = nlf_get_u8 },
691 { .off_in = _IN(ifa_scope), .off_out = _OUT(ifa_scope), .cb = nlf_get_u8 },
692 { .off_in = _IN(ifa_flags), .off_out = _OUT(ifa_flags), .cb = nlf_get_u8_u32 },
693 { .off_in = _IN(ifa_index), .off_out = _OUT(ifa_index), .cb = nlf_get_u32 },
696 static const struct nlattr_parser nla_p_ifa[] = {
697 { .type = IFA_ADDRESS, .off = _OUT(ifa_address), .cb = nlattr_get_ip },
698 { .type = IFA_LOCAL, .off = _OUT(ifa_local), .cb = nlattr_get_ip },
699 { .type = IFA_FLAGS, .off = _OUT(ifa_flags), .cb = nlattr_get_uint32 },
703 NL_DECLARE_PARSER(ifaddrmsg_parser, struct ifaddrmsg, nlf_p_ifa, nla_p_ifa);
708 {ifa_family=AF_INET, ifa_prefixlen=8, ifa_flags=IFA_F_PERMANENT, ifa_scope=RT_SCOPE_HOST, ifa_index=if_nametoindex("lo")},
710 {{nla_len=8, nla_type=IFA_ADDRESS}, inet_addr("127.0.0.1")},
711 {{nla_len=8, nla_type=IFA_LOCAL}, inet_addr("127.0.0.1")},
712 {{nla_len=7, nla_type=IFA_LABEL}, "lo"},
713 {{nla_len=8, nla_type=IFA_FLAGS}, IFA_F_PERMANENT},
714 {{nla_len=20, nla_type=IFA_CACHEINFO}, {ifa_prefered=4294967295, ifa_valid=4294967295, cstamp=3619, tstamp=3619}}]},
717 {{len=72, type=RTM_NEWADDR, flags=NLM_F_MULTI, seq=1642191126, pid=566735},
718 {ifa_family=AF_INET6, ifa_prefixlen=96, ifa_flags=IFA_F_PERMANENT, ifa_scope=RT_SCOPE_UNIVERSE, ifa_index=if_nametoindex("virbr0")},
720 {{nla_len=20, nla_type=IFA_ADDRESS}, inet_pton(AF_INET6, "2a01:4f8:13a:70c:ffff::1")},
721 {{nla_len=20, nla_type=IFA_CACHEINFO}, {ifa_prefered=4294967295, ifa_valid=4294967295, cstamp=4283, tstamp=4283}},
722 {{nla_len=8, nla_type=IFA_FLAGS}, IFA_F_PERMANENT}]},
726 ifa_get_scope(const struct ifaddr *ifa)
728 const struct sockaddr *sa;
729 uint8_t addr_scope = RT_SCOPE_UNIVERSE;
732 switch (sa->sa_family) {
737 addr = ((const struct sockaddr_in *)sa)->sin_addr;
738 if (IN_LOOPBACK(addr.s_addr))
739 addr_scope = RT_SCOPE_HOST;
740 else if (IN_LINKLOCAL(addr.s_addr))
741 addr_scope = RT_SCOPE_LINK;
748 const struct in6_addr *addr;
749 addr = &((const struct sockaddr_in6 *)sa)->sin6_addr;
750 if (IN6_IS_ADDR_LOOPBACK(addr))
751 addr_scope = RT_SCOPE_HOST;
752 else if (IN6_IS_ADDR_LINKLOCAL(addr))
753 addr_scope = RT_SCOPE_LINK;
763 inet6_get_plen(const struct in6_addr *addr)
766 return (bitcount32(addr->s6_addr32[0]) + bitcount32(addr->s6_addr32[1]) +
767 bitcount32(addr->s6_addr32[2]) + bitcount32(addr->s6_addr32[3]));
771 get_sa_plen(const struct sockaddr *sa)
774 const struct in_addr *paddr;
777 const struct in6_addr *paddr6;
780 switch (sa->sa_family) {
785 paddr = &(((const struct sockaddr_in *)sa)->sin_addr);
786 return bitcount32(paddr->s_addr);;
792 paddr6 = &(((const struct sockaddr_in6 *)sa)->sin6_addr);
793 return inet6_get_plen(paddr6);
802 * {'attrs': [('IFA_ADDRESS', '12.0.0.1'),
803 ('IFA_LOCAL', '12.0.0.1'),
804 ('IFA_LABEL', 'eth10'),
806 ('IFA_CACHEINFO', {'ifa_preferred': 4294967295, 'ifa_valid': 4294967295, 'cstamp': 63745746, 'tstamp': 63745746})],
809 dump_iface_addr(struct nl_writer *nw, struct ifnet *ifp, struct ifaddr *ifa,
810 const struct nlmsghdr *hdr)
812 struct ifaddrmsg *ifamsg;
813 struct sockaddr *sa = ifa->ifa_addr;
815 NL_LOG(LOG_DEBUG3, "dumping ifa %p type %s(%d) for interface %s",
816 ifa, rib_print_family(sa->sa_family), sa->sa_family, if_name(ifp));
818 if (!nlmsg_reply(nw, hdr, sizeof(struct ifaddrmsg)))
821 ifamsg = nlmsg_reserve_object(nw, struct ifaddrmsg);
822 ifamsg->ifa_family = sa->sa_family;
823 ifamsg->ifa_prefixlen = get_sa_plen(ifa->ifa_netmask);
824 ifamsg->ifa_flags = 0; // ifa_flags is useless
825 ifamsg->ifa_scope = ifa_get_scope(ifa);
826 ifamsg->ifa_index = ifp->if_index;
828 if (ifp->if_flags & IFF_POINTOPOINT) {
829 dump_sa(nw, IFA_ADDRESS, ifa->ifa_dstaddr);
830 dump_sa(nw, IFA_LOCAL, sa);
832 dump_sa(nw, IFA_ADDRESS, sa);
835 * In most cases, IFA_ADDRESS == IFA_LOCAL
836 * Skip IFA_LOCAL for anything except INET
838 if (sa->sa_family == AF_INET)
839 dump_sa(nw, IFA_LOCAL, sa);
842 if (ifp->if_flags & IFF_BROADCAST)
843 dump_sa(nw, IFA_BROADCAST, ifa->ifa_broadaddr);
845 nlattr_add_string(nw, IFA_LABEL, if_name(ifp));
847 uint32_t val = 0; // ifa->ifa_flags;
848 nlattr_add_u32(nw, IFA_FLAGS, val);
853 NL_LOG(LOG_DEBUG, "Failed to dump ifa type %s(%d) for interface %s",
854 rib_print_family(sa->sa_family), sa->sa_family, if_name(ifp));
860 dump_iface_addrs(struct netlink_walkargs *wa, struct ifnet *ifp)
864 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
865 if (wa->family != 0 && wa->family != ifa->ifa_addr->sa_family)
867 if (ifa->ifa_addr->sa_family == AF_LINK)
870 if (!dump_iface_addr(wa->nw, ifp, ifa, &wa->hdr))
879 rtnl_handle_getaddr(struct nlmsghdr *hdr, struct nlpcb *nlp, struct nl_pstate *npt)
884 struct nl_parsed_ifa attrs = {};
885 error = nl_parse_nlmsg(hdr, &ifaddrmsg_parser, npt, &attrs);
889 struct netlink_walkargs wa = {
892 .family = attrs.ifa_family,
893 .hdr.nlmsg_pid = hdr->nlmsg_pid,
894 .hdr.nlmsg_seq = hdr->nlmsg_seq,
895 .hdr.nlmsg_flags = hdr->nlmsg_flags | NLM_F_MULTI,
896 .hdr.nlmsg_type = NL_RTM_NEWADDR,
899 NL_LOG(LOG_DEBUG2, "Start dump");
901 if (attrs.ifa_index != 0) {
902 ifp = ifnet_byindex(attrs.ifa_index);
906 error = dump_iface_addrs(&wa, ifp);
908 CK_STAILQ_FOREACH(ifp, &V_ifnet, if_link) {
909 error = dump_iface_addrs(&wa, ifp);
915 NL_LOG(LOG_DEBUG2, "End dump, iterated %d dumped %d", wa.count, wa.dumped);
917 if (!nlmsg_end_dump(wa.nw, error, &wa.hdr)) {
918 NL_LOG(LOG_DEBUG, "Unable to finalize the dump");
926 rtnl_handle_ifaddr(void *arg __unused, struct ifaddr *ifa, int cmd)
928 struct nlmsghdr hdr = {};
929 struct nl_writer nw = {};
932 switch (ifa->ifa_addr->sa_family) {
935 group = RTNLGRP_IPV4_IFADDR;
940 group = RTNLGRP_IPV6_IFADDR;
944 NL_LOG(LOG_DEBUG2, "ifa notification for unknown AF: %d",
945 ifa->ifa_addr->sa_family);
949 if (!nl_has_listeners(NETLINK_ROUTE, group))
952 if (!nlmsg_get_group_writer(&nw, NLMSG_LARGE, NETLINK_ROUTE, group)) {
953 NL_LOG(LOG_DEBUG, "error allocating group writer");
957 hdr.nlmsg_type = (cmd == RTM_DELETE) ? NL_RTM_DELADDR : NL_RTM_NEWADDR;
959 dump_iface_addr(&nw, ifa->ifa_ifp, ifa, &hdr);
964 rtnl_handle_ifevent(struct ifnet *ifp, int nlmsg_type, int if_flags_mask)
966 struct nlmsghdr hdr = { .nlmsg_type = nlmsg_type };
967 struct nl_writer nw = {};
969 if (!nl_has_listeners(NETLINK_ROUTE, RTNLGRP_LINK))
972 if (!nlmsg_get_group_writer(&nw, NLMSG_LARGE, NETLINK_ROUTE, RTNLGRP_LINK)) {
973 NL_LOG(LOG_DEBUG, "error allocating mbuf");
976 dump_iface(&nw, ifp, &hdr, if_flags_mask);
981 rtnl_handle_ifattach(void *arg, struct ifnet *ifp)
983 NL_LOG(LOG_DEBUG2, "ifnet %s", if_name(ifp));
984 rtnl_handle_ifevent(ifp, NL_RTM_NEWLINK, 0);
988 rtnl_handle_ifdetach(void *arg, struct ifnet *ifp)
990 NL_LOG(LOG_DEBUG2, "ifnet %s", if_name(ifp));
991 rtnl_handle_ifevent(ifp, NL_RTM_DELLINK, 0);
995 rtnl_handle_iflink(void *arg, struct ifnet *ifp)
997 NL_LOG(LOG_DEBUG2, "ifnet %s", if_name(ifp));
998 rtnl_handle_ifevent(ifp, NL_RTM_NEWLINK, 0);
1002 rtnl_handle_ifnet_event(struct ifnet *ifp, int if_flags_mask)
1004 NL_LOG(LOG_DEBUG2, "ifnet %s", if_name(ifp));
1005 rtnl_handle_ifevent(ifp, NL_RTM_NEWLINK, if_flags_mask);
1008 static const struct rtnl_cmd_handler cmd_handlers[] = {
1010 .cmd = NL_RTM_GETLINK,
1011 .name = "RTM_GETLINK",
1012 .cb = &rtnl_handle_getlink,
1013 .flags = RTNL_F_NOEPOCH,
1016 .cmd = NL_RTM_DELLINK,
1017 .name = "RTM_DELLINK",
1018 .cb = &rtnl_handle_dellink,
1019 .priv = PRIV_NET_IFDESTROY,
1020 .flags = RTNL_F_NOEPOCH,
1023 .cmd = NL_RTM_NEWLINK,
1024 .name = "RTM_NEWLINK",
1025 .cb = &rtnl_handle_newlink,
1026 .priv = PRIV_NET_IFCREATE,
1027 .flags = RTNL_F_NOEPOCH,
1030 .cmd = NL_RTM_GETADDR,
1031 .name = "RTM_GETADDR",
1032 .cb = &rtnl_handle_getaddr,
1035 .cmd = NL_RTM_NEWADDR,
1036 .name = "RTM_NEWADDR",
1037 .cb = &rtnl_handle_getaddr,
1040 .cmd = NL_RTM_DELADDR,
1041 .name = "RTM_DELADDR",
1042 .cb = &rtnl_handle_getaddr,
1046 static const struct nlhdr_parser *all_parsers[] = { &ifmsg_parser, &ifaddrmsg_parser };
1049 rtnl_iface_add_cloner(struct nl_cloner *cloner)
1051 sx_xlock(&rtnl_cloner_lock);
1052 SLIST_INSERT_HEAD(&nl_cloners, cloner, next);
1053 sx_xunlock(&rtnl_cloner_lock);
1057 rtnl_iface_del_cloner(struct nl_cloner *cloner)
1059 sx_xlock(&rtnl_cloner_lock);
1060 SLIST_REMOVE(&nl_cloners, cloner, nl_cloner, next);
1061 sx_xunlock(&rtnl_cloner_lock);
1064 static struct nl_cloner *
1065 rtnl_iface_find_cloner_locked(const char *name)
1067 struct nl_cloner *cloner;
1069 SLIST_FOREACH(cloner, &nl_cloners, next) {
1070 if (!strcmp(name, cloner->name))
1078 rtnl_ifaces_init(void)
1080 ifattach_event = EVENTHANDLER_REGISTER(
1081 ifnet_arrival_event, rtnl_handle_ifattach, NULL,
1082 EVENTHANDLER_PRI_ANY);
1083 ifdetach_event = EVENTHANDLER_REGISTER(
1084 ifnet_departure_event, rtnl_handle_ifdetach, NULL,
1085 EVENTHANDLER_PRI_ANY);
1086 ifaddr_event = EVENTHANDLER_REGISTER(
1087 rt_addrmsg, rtnl_handle_ifaddr, NULL,
1088 EVENTHANDLER_PRI_ANY);
1089 iflink_event = EVENTHANDLER_REGISTER(
1090 ifnet_link_event, rtnl_handle_iflink, NULL,
1091 EVENTHANDLER_PRI_ANY);
1092 NL_VERIFY_PARSERS(all_parsers);
1093 rtnl_iface_drivers_register();
1094 rtnl_register_messages(cmd_handlers, NL_ARRAY_LEN(cmd_handlers));
1098 rtnl_ifaces_destroy(void)
1100 EVENTHANDLER_DEREGISTER(ifnet_arrival_event, ifattach_event);
1101 EVENTHANDLER_DEREGISTER(ifnet_departure_event, ifdetach_event);
1102 EVENTHANDLER_DEREGISTER(rt_addrmsg, ifaddr_event);
1103 EVENTHANDLER_DEREGISTER(ifnet_link_event, iflink_event);