2 * Code for encoding/decoding FPM messages that are in netlink format.
4 * Copyright (C) 1997, 98, 99 Kunihiro Ishiguro
5 * Copyright (C) 2012 by Open Source Routing.
6 * Copyright (C) 2012 by Internet Systems Consortium, Inc. ("ISC")
8 * This file is part of GNU Zebra.
10 * GNU Zebra is free software; you can redistribute it and/or modify it
11 * under the terms of the GNU General Public License as published by the
12 * Free Software Foundation; either version 2, or (at your option) any
15 * GNU Zebra is distributed in the hope that it will be useful, but
16 * WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * General Public License for more details.
20 * You should have received a copy of the GNU General Public License along
21 * with this program; see the file COPYING; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
34 #include "zebra/zserv.h"
35 #include "zebra/zebra_router.h"
36 #include "zebra/zebra_dplane.h"
37 #include "zebra/zebra_ns.h"
38 #include "zebra/zebra_vrf.h"
39 #include "zebra/kernel_netlink.h"
40 #include "zebra/rt_netlink.h"
43 #include "zebra/zebra_fpm_private.h"
44 #include "zebra/zebra_vxlan_private.h"
45 #include "zebra/interface.h"
50 * The size of an address in a given address family.
52 static size_t af_addr_size(uint8_t af
)
67 * We plan to use RTA_ENCAP_TYPE attribute for VxLAN encap as well.
68 * Currently, values 0 to 8 for this attribute are used by lwtunnel_encap_types
69 * So, we cannot use these values for VxLAN encap.
71 enum fpm_nh_encap_type_t
{
72 FPM_NH_ENCAP_NONE
= 0,
73 FPM_NH_ENCAP_VXLAN
= 100,
78 * fpm_nh_encap_type_to_str
80 static const char *fpm_nh_encap_type_to_str(enum fpm_nh_encap_type_t encap_type
)
83 case FPM_NH_ENCAP_NONE
:
86 case FPM_NH_ENCAP_VXLAN
:
89 case FPM_NH_ENCAP_MAX
:
96 struct vxlan_encap_info_t
{
100 enum vxlan_encap_info_type_t
{
104 struct fpm_nh_encap_info_t
{
105 enum fpm_nh_encap_type_t encap_type
;
107 struct vxlan_encap_info_t vxlan_encap
;
114 * Holds information about a single nexthop for netlink. These info
115 * structures are transient and may contain pointers into rib
116 * data structures for convenience.
118 struct netlink_nh_info
{
120 union g_addr
*gateway
;
123 * Information from the struct nexthop from which this nh was
124 * derived. For debug purposes only.
127 enum nexthop_types_t type
;
128 struct fpm_nh_encap_info_t encap_info
;
134 * A structure for holding information for a netlink route message.
136 struct netlink_route_info
{
141 uint8_t rtm_protocol
;
143 struct prefix
*prefix
;
145 unsigned int num_nhs
;
150 struct netlink_nh_info nhs
[MULTIPATH_NUM
];
151 union g_addr
*pref_src
;
155 * netlink_route_info_add_nh
157 * Add information about the given nexthop to the given route info
160 * Returns true if a nexthop was added, false otherwise.
162 static int netlink_route_info_add_nh(struct netlink_route_info
*ri
,
163 struct nexthop
*nexthop
,
164 struct route_entry
*re
)
166 struct netlink_nh_info nhi
;
168 struct zebra_vrf
*zvrf
= NULL
;
169 struct interface
*ifp
= NULL
, *link_if
= NULL
;
170 struct zebra_if
*zif
= NULL
;
173 memset(&nhi
, 0, sizeof(nhi
));
176 if (ri
->num_nhs
>= (int)array_size(ri
->nhs
))
179 nhi
.recursive
= nexthop
->rparent
? 1 : 0;
180 nhi
.type
= nexthop
->type
;
181 nhi
.if_index
= nexthop
->ifindex
;
183 if (nexthop
->type
== NEXTHOP_TYPE_IPV4
184 || nexthop
->type
== NEXTHOP_TYPE_IPV4_IFINDEX
) {
185 nhi
.gateway
= &nexthop
->gate
;
186 if (nexthop
->src
.ipv4
.s_addr
!= INADDR_ANY
)
190 if (nexthop
->type
== NEXTHOP_TYPE_IPV6
191 || nexthop
->type
== NEXTHOP_TYPE_IPV6_IFINDEX
) {
192 /* Special handling for IPv4 route with IPv6 Link Local next hop
194 if (ri
->af
== AF_INET
)
195 nhi
.gateway
= &ipv4ll_gateway
;
197 nhi
.gateway
= &nexthop
->gate
;
200 if (nexthop
->type
== NEXTHOP_TYPE_IFINDEX
) {
201 if (nexthop
->src
.ipv4
.s_addr
!= INADDR_ANY
)
205 if (!nhi
.gateway
&& nhi
.if_index
== 0)
208 if (re
&& CHECK_FLAG(re
->flags
, ZEBRA_FLAG_EVPN_ROUTE
)) {
209 nhi
.encap_info
.encap_type
= FPM_NH_ENCAP_VXLAN
;
211 /* Extract VNI id for the nexthop SVI interface */
212 zvrf
= zebra_vrf_lookup_by_id(nexthop
->vrf_id
);
214 ifp
= if_lookup_by_index_per_ns(zvrf
->zns
,
217 zif
= (struct zebra_if
*)ifp
->info
;
219 if (IS_ZEBRA_IF_BRIDGE(ifp
))
221 else if (IS_ZEBRA_IF_VLAN(ifp
))
223 if_lookup_by_index_per_ns(
227 vni
= vni_id_from_svi(ifp
,
233 nhi
.encap_info
.vxlan_encap
.vni
= vni
;
237 * We have a valid nhi. Copy the structure over to the route_info.
239 ri
->nhs
[ri
->num_nhs
] = nhi
;
242 if (src
&& !ri
->pref_src
)
249 * netlink_proto_from_route_type
251 static uint8_t netlink_proto_from_route_type(int type
)
254 case ZEBRA_ROUTE_KERNEL
:
255 case ZEBRA_ROUTE_CONNECT
:
256 return RTPROT_KERNEL
;
264 * netlink_route_info_fill
266 * Fill out the route information object from the given route.
268 * Returns true on success and false on failure.
270 static int netlink_route_info_fill(struct netlink_route_info
*ri
, int cmd
,
271 rib_dest_t
*dest
, struct route_entry
*re
)
273 struct nexthop
*nexthop
;
274 struct rib_table_info
*table_info
=
275 rib_table_info(rib_dest_table(dest
));
276 struct zebra_vrf
*zvrf
= table_info
->zvrf
;
278 memset(ri
, 0, sizeof(*ri
));
280 ri
->prefix
= rib_dest_prefix(dest
);
281 ri
->af
= rib_dest_af(dest
);
283 if (zvrf
&& zvrf
->zns
)
284 ri
->nlmsg_pid
= zvrf
->zns
->netlink_dplane
.snl
.nl_pid
;
286 ri
->nlmsg_type
= cmd
;
287 ri
->rtm_table
= table_info
->table_id
;
288 ri
->rtm_protocol
= RTPROT_UNSPEC
;
291 * An RTM_DELROUTE need not be accompanied by any nexthops,
292 * particularly in our communication with the FPM.
294 if (cmd
== RTM_DELROUTE
&& !re
)
298 zfpm_debug("%s: Expected non-NULL re pointer", __func__
);
302 ri
->rtm_protocol
= netlink_proto_from_route_type(re
->type
);
303 ri
->rtm_type
= RTN_UNICAST
;
304 ri
->metric
= &re
->metric
;
306 for (ALL_NEXTHOPS(re
->nhe
->nhg
, nexthop
)) {
307 if (ri
->num_nhs
>= zrouter
.multipath_num
)
310 if (CHECK_FLAG(nexthop
->flags
, NEXTHOP_FLAG_RECURSIVE
))
313 if (nexthop
->type
== NEXTHOP_TYPE_BLACKHOLE
) {
314 switch (nexthop
->bh_type
) {
315 case BLACKHOLE_ADMINPROHIB
:
316 ri
->rtm_type
= RTN_PROHIBIT
;
318 case BLACKHOLE_REJECT
:
319 ri
->rtm_type
= RTN_UNREACHABLE
;
323 ri
->rtm_type
= RTN_BLACKHOLE
;
328 if ((cmd
== RTM_NEWROUTE
329 && CHECK_FLAG(nexthop
->flags
, NEXTHOP_FLAG_ACTIVE
))
330 || (cmd
== RTM_DELROUTE
331 && CHECK_FLAG(re
->status
, ROUTE_ENTRY_INSTALLED
))) {
332 netlink_route_info_add_nh(ri
, nexthop
, re
);
336 if (ri
->num_nhs
== 0) {
337 switch (ri
->rtm_type
) {
339 case RTN_UNREACHABLE
:
343 /* If there is no useful nexthop then return. */
345 "netlink_encode_route(): No useful nexthop.");
354 * netlink_route_info_encode
356 * Returns the number of bytes written to the buffer. 0 or a negative
357 * value indicates an error.
359 static int netlink_route_info_encode(struct netlink_route_info
*ri
,
360 char *in_buf
, size_t in_buf_len
)
363 unsigned int nexthop_num
= 0;
365 struct netlink_nh_info
*nhi
;
366 enum fpm_nh_encap_type_t encap
;
367 struct rtattr
*nest
, *inner_nest
;
368 struct rtnexthop
*rtnh
;
369 struct vxlan_encap_info_t
*vxlan
;
370 struct in6_addr ipv6
;
378 req
= (void *)in_buf
;
380 buf_offset
= ((char *)req
->buf
) - ((char *)req
);
382 if (in_buf_len
< buf_offset
) {
387 memset(req
, 0, buf_offset
);
389 bytelen
= af_addr_size(ri
->af
);
391 req
->n
.nlmsg_len
= NLMSG_LENGTH(sizeof(struct rtmsg
));
392 req
->n
.nlmsg_flags
= NLM_F_CREATE
| NLM_F_REQUEST
;
393 req
->n
.nlmsg_pid
= ri
->nlmsg_pid
;
394 req
->n
.nlmsg_type
= ri
->nlmsg_type
;
395 req
->r
.rtm_family
= ri
->af
;
398 * rtm_table field is a uchar field which can accomodate table_id less
400 * To support table id greater than 255, if the table_id is greater than
401 * 255, set rtm_table to RT_TABLE_UNSPEC and add RTA_TABLE attribute
402 * with 32 bit value as the table_id.
404 if (ri
->rtm_table
< 256)
405 req
->r
.rtm_table
= ri
->rtm_table
;
407 req
->r
.rtm_table
= RT_TABLE_UNSPEC
;
408 nl_attr_put32(&req
->n
, in_buf_len
, RTA_TABLE
, ri
->rtm_table
);
411 req
->r
.rtm_dst_len
= ri
->prefix
->prefixlen
;
412 req
->r
.rtm_protocol
= ri
->rtm_protocol
;
413 req
->r
.rtm_scope
= RT_SCOPE_UNIVERSE
;
415 nl_attr_put(&req
->n
, in_buf_len
, RTA_DST
, &ri
->prefix
->u
.prefix
,
418 req
->r
.rtm_type
= ri
->rtm_type
;
422 nl_attr_put32(&req
->n
, in_buf_len
, RTA_PRIORITY
, *ri
->metric
);
424 if (ri
->num_nhs
== 0)
427 if (ri
->num_nhs
== 1) {
431 if (nhi
->type
== NEXTHOP_TYPE_IPV4_IFINDEX
432 && ri
->af
== AF_INET6
) {
433 ipv4_to_ipv4_mapped_ipv6(&ipv6
,
435 nl_attr_put(&req
->n
, in_buf_len
, RTA_GATEWAY
,
438 nl_attr_put(&req
->n
, in_buf_len
, RTA_GATEWAY
,
439 nhi
->gateway
, bytelen
);
443 nl_attr_put32(&req
->n
, in_buf_len
, RTA_OIF
,
447 encap
= nhi
->encap_info
.encap_type
;
449 case FPM_NH_ENCAP_NONE
:
450 case FPM_NH_ENCAP_MAX
:
452 case FPM_NH_ENCAP_VXLAN
:
453 nl_attr_put16(&req
->n
, in_buf_len
, RTA_ENCAP_TYPE
,
455 vxlan
= &nhi
->encap_info
.vxlan_encap
;
456 nest
= nl_attr_nest(&req
->n
, in_buf_len
, RTA_ENCAP
);
457 nl_attr_put32(&req
->n
, in_buf_len
, VXLAN_VNI
,
459 nl_attr_nest_end(&req
->n
, nest
);
469 nest
= nl_attr_nest(&req
->n
, in_buf_len
, RTA_MULTIPATH
);
471 for (nexthop_num
= 0; nexthop_num
< ri
->num_nhs
; nexthop_num
++) {
472 rtnh
= nl_attr_rtnh(&req
->n
, in_buf_len
);
473 nhi
= &ri
->nhs
[nexthop_num
];
476 nl_attr_put(&req
->n
, in_buf_len
, RTA_GATEWAY
,
477 nhi
->gateway
, bytelen
);
480 rtnh
->rtnh_ifindex
= nhi
->if_index
;
483 encap
= nhi
->encap_info
.encap_type
;
485 case FPM_NH_ENCAP_NONE
:
486 case FPM_NH_ENCAP_MAX
:
488 case FPM_NH_ENCAP_VXLAN
:
489 nl_attr_put16(&req
->n
, in_buf_len
, RTA_ENCAP_TYPE
,
491 vxlan
= &nhi
->encap_info
.vxlan_encap
;
493 nl_attr_nest(&req
->n
, in_buf_len
, RTA_ENCAP
);
494 nl_attr_put32(&req
->n
, in_buf_len
, VXLAN_VNI
,
496 nl_attr_nest_end(&req
->n
, inner_nest
);
500 nl_attr_rtnh_end(&req
->n
, rtnh
);
503 nl_attr_nest_end(&req
->n
, nest
);
504 assert(nest
->rta_len
> RTA_LENGTH(0));
509 nl_attr_put(&req
->n
, in_buf_len
, RTA_PREFSRC
, &ri
->pref_src
,
513 assert(req
->n
.nlmsg_len
< in_buf_len
);
514 return req
->n
.nlmsg_len
;
518 * zfpm_log_route_info
520 * Helper function to log the information in a route_info structure.
522 static void zfpm_log_route_info(struct netlink_route_info
*ri
,
525 struct netlink_nh_info
*nhi
;
527 char buf
[PREFIX_STRLEN
];
529 zfpm_debug("%s : %s %pFX, Proto: %s, Metric: %u", label
,
530 nl_msg_type_to_str(ri
->nlmsg_type
), ri
->prefix
,
531 nl_rtproto_to_str(ri
->rtm_protocol
),
532 ri
->metric
? *ri
->metric
: 0);
534 for (i
= 0; i
< ri
->num_nhs
; i
++) {
537 if (ri
->af
== AF_INET
)
538 inet_ntop(AF_INET
, &nhi
->gateway
, buf
, sizeof(buf
));
540 inet_ntop(AF_INET6
, &nhi
->gateway
, buf
, sizeof(buf
));
542 zfpm_debug(" Intf: %u, Gateway: %s, Recursive: %s, Type: %s, Encap type: %s",
543 nhi
->if_index
, buf
, nhi
->recursive
? "yes" : "no",
544 nexthop_type_to_str(nhi
->type
),
545 fpm_nh_encap_type_to_str(nhi
->encap_info
.encap_type
)
551 * zfpm_netlink_encode_route
553 * Create a netlink message corresponding to the given route in the
554 * given buffer space.
556 * Returns the number of bytes written to the buffer. 0 or a negative
557 * value indicates an error.
559 int zfpm_netlink_encode_route(int cmd
, rib_dest_t
*dest
, struct route_entry
*re
,
560 char *in_buf
, size_t in_buf_len
)
562 struct netlink_route_info ri_space
, *ri
;
566 if (!netlink_route_info_fill(ri
, cmd
, dest
, re
))
569 zfpm_log_route_info(ri
, __func__
);
571 return netlink_route_info_encode(ri
, in_buf
, in_buf_len
);
575 * zfpm_netlink_encode_mac
577 * Create a netlink message corresponding to the given MAC.
579 * Returns the number of bytes written to the buffer. 0 or a negative
580 * value indicates an error.
582 int zfpm_netlink_encode_mac(struct fpm_mac_info_t
*mac
, char *in_buf
,
592 req
= (void *)in_buf
;
594 buf_offset
= offsetof(struct macmsg
, buf
);
595 if (in_buf_len
< buf_offset
)
597 memset(req
, 0, buf_offset
);
599 /* Construct nlmsg header */
600 req
->hdr
.nlmsg_len
= NLMSG_LENGTH(sizeof(struct ndmsg
));
601 req
->hdr
.nlmsg_type
= CHECK_FLAG(mac
->fpm_flags
, ZEBRA_MAC_DELETE_FPM
) ?
602 RTM_DELNEIGH
: RTM_NEWNEIGH
;
603 req
->hdr
.nlmsg_flags
= NLM_F_REQUEST
;
604 if (req
->hdr
.nlmsg_type
== RTM_NEWNEIGH
)
605 req
->hdr
.nlmsg_flags
|= (NLM_F_CREATE
| NLM_F_REPLACE
);
607 /* Construct ndmsg */
608 req
->ndm
.ndm_family
= AF_BRIDGE
;
609 req
->ndm
.ndm_ifindex
= mac
->vxlan_if
;
611 req
->ndm
.ndm_state
= NUD_REACHABLE
;
612 req
->ndm
.ndm_flags
|= NTF_SELF
| NTF_MASTER
;
613 if (CHECK_FLAG(mac
->zebra_flags
,
614 (ZEBRA_MAC_STICKY
| ZEBRA_MAC_REMOTE_DEF_GW
)))
615 req
->ndm
.ndm_state
|= NUD_NOARP
;
617 req
->ndm
.ndm_flags
|= NTF_EXT_LEARNED
;
620 nl_attr_put(&req
->hdr
, in_buf_len
, NDA_LLADDR
, &mac
->macaddr
, 6);
621 nl_attr_put(&req
->hdr
, in_buf_len
, NDA_DST
, &mac
->r_vtep_ip
, 4);
622 nl_attr_put32(&req
->hdr
, in_buf_len
, NDA_MASTER
, mac
->svi_if
);
623 nl_attr_put32(&req
->hdr
, in_buf_len
, NDA_VNI
, mac
->vni
);
625 assert(req
->hdr
.nlmsg_len
< in_buf_len
);
627 zfpm_debug("Tx %s family %s ifindex %u MAC %pEA DEST %pI4",
628 nl_msg_type_to_str(req
->hdr
.nlmsg_type
),
629 nl_family_to_str(req
->ndm
.ndm_family
), req
->ndm
.ndm_ifindex
,
630 &mac
->macaddr
, &mac
->r_vtep_ip
);
632 return req
->hdr
.nlmsg_len
;
635 #endif /* HAVE_NETLINK */