2 * pim_bsm.c: PIM BSM handling routines
4 * Copyright (C) 2018-19 Vmware, Inc.
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
12 * This program is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with this program; see the file COPYING; if not, write to the
19 * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston,
29 #include "pim_iface.h"
30 #include "pim_instance.h"
32 #include "pim_hello.h"
38 /* Functions forward declaration */
39 static void pim_bs_timer_start(struct bsm_scope
*scope
, int bs_timeout
);
40 static void pim_g2rp_timer_start(struct bsm_rpinfo
*bsrp
, int hold_time
);
41 static inline void pim_g2rp_timer_restart(struct bsm_rpinfo
*bsrp
,
45 DEFINE_MTYPE_STATIC(PIMD
, PIM_BSGRP_NODE
, "PIM BSR advertised grp info")
46 DEFINE_MTYPE_STATIC(PIMD
, PIM_BSRP_NODE
, "PIM BSR advertised RP info")
47 DEFINE_MTYPE_STATIC(PIMD
, PIM_BSM_INFO
, "PIM BSM Info")
48 DEFINE_MTYPE_STATIC(PIMD
, PIM_BSM_PKT_VAR_MEM
, "PIM BSM Packet")
50 /* All bsm packets forwarded shall be fit within ip mtu less iphdr(max) */
51 #define MAX_IP_HDR_LEN 24
53 /* pim_bsm_write_config - Write the interface pim bsm configuration.*/
54 void pim_bsm_write_config(struct vty
*vty
, struct interface
*ifp
)
56 struct pim_interface
*pim_ifp
= ifp
->info
;
59 if (!pim_ifp
->bsm_enable
)
60 vty_out(vty
, " no ip pim bsm\n");
61 if (!pim_ifp
->ucast_bsm_accept
)
62 vty_out(vty
, " no ip pim unicast-bsm\n");
66 static void pim_free_bsgrp_data(struct bsgrp_node
*bsgrp_node
)
68 if (bsgrp_node
->bsrp_list
)
69 list_delete(&bsgrp_node
->bsrp_list
);
70 if (bsgrp_node
->partial_bsrp_list
)
71 list_delete(&bsgrp_node
->partial_bsrp_list
);
72 XFREE(MTYPE_PIM_BSGRP_NODE
, bsgrp_node
);
75 static void pim_free_bsgrp_node(struct route_table
*rt
, struct prefix
*grp
)
77 struct route_node
*rn
;
79 rn
= route_node_lookup(rt
, grp
);
82 route_unlock_node(rn
);
83 route_unlock_node(rn
);
87 static void pim_bsm_node_free(struct bsm_info
*bsm
)
89 XFREE(MTYPE_PIM_BSM_PKT_VAR_MEM
, bsm
->bsm
);
90 XFREE(MTYPE_PIM_BSM_INFO
, bsm
);
93 static int pim_g2rp_list_compare(struct bsm_rpinfo
*node1
,
94 struct bsm_rpinfo
*node2
)
97 * Step-1 : Loweset Rp priority will have higher precedance.
98 * Step-2 : If priority same then higher hash val will have
100 * Step-3 : If Hash val is same then highest rp address will
103 if (node1
->rp_prio
< node2
->rp_prio
)
105 if (node1
->rp_prio
> node2
->rp_prio
)
107 if (node1
->hash
< node2
->hash
)
109 if (node1
->hash
> node2
->hash
)
111 if (node1
->rp_address
.s_addr
< node2
->rp_address
.s_addr
)
113 if (node1
->rp_address
.s_addr
> node2
->rp_address
.s_addr
)
118 static void pim_free_bsrp_node(struct bsm_rpinfo
*bsrp_info
)
120 if (bsrp_info
->g2rp_timer
)
121 THREAD_OFF(bsrp_info
->g2rp_timer
);
122 XFREE(MTYPE_PIM_BSRP_NODE
, bsrp_info
);
125 static struct list
*pim_alloc_bsrp_list(void)
127 struct list
*new_list
= NULL
;
129 new_list
= list_new();
134 new_list
->cmp
= (int (*)(void *, void *))pim_g2rp_list_compare
;
135 new_list
->del
= (void (*)(void *))pim_free_bsrp_node
;
140 static struct bsgrp_node
*pim_bsm_new_bsgrp_node(struct route_table
*rt
,
143 struct route_node
*rn
;
144 struct bsgrp_node
*bsgrp
;
146 rn
= route_node_get(rt
, grp
);
148 zlog_warn("%s: route node creation failed", __func__
);
151 bsgrp
= XCALLOC(MTYPE_PIM_BSGRP_NODE
, sizeof(struct bsgrp_node
));
154 bsgrp
->bsrp_list
= pim_alloc_bsrp_list();
155 bsgrp
->partial_bsrp_list
= pim_alloc_bsrp_list();
157 if ((!bsgrp
->bsrp_list
) || (!bsgrp
->partial_bsrp_list
)) {
158 route_unlock_node(rn
);
159 pim_free_bsgrp_data(bsgrp
);
163 prefix_copy(&bsgrp
->group
, grp
);
167 static int pim_on_bs_timer(struct thread
*t
)
169 struct route_node
*rn
;
170 struct bsm_scope
*scope
;
171 struct bsgrp_node
*bsgrp_node
;
172 struct bsm_rpinfo
*bsrp
;
174 char buf
[PREFIX2STR_BUFFER
];
175 bool is_bsr_tracking
= true;
177 scope
= THREAD_ARG(t
);
178 THREAD_OFF(scope
->bs_timer
);
181 zlog_debug("%s: Bootstrap Timer expired for scope: %d",
182 __func__
, scope
->sz_id
);
184 /* Remove next hop tracking for the bsr */
185 nht_p
.family
= AF_INET
;
186 nht_p
.prefixlen
= IPV4_MAX_BITLEN
;
187 nht_p
.u
.prefix4
= scope
->current_bsr
;
189 prefix2str(&nht_p
, buf
, sizeof(buf
));
190 zlog_debug("%s: Deregister BSR addr %s with Zebra NHT",
193 pim_delete_tracked_nexthop(scope
->pim
, &nht_p
, NULL
, NULL
,
196 /* Reset scope zone data */
197 scope
->accept_nofwd_bsm
= false;
198 scope
->state
= ACCEPT_ANY
;
199 scope
->current_bsr
.s_addr
= INADDR_ANY
;
200 scope
->current_bsr_prio
= 0;
201 scope
->current_bsr_first_ts
= 0;
202 scope
->current_bsr_last_ts
= 0;
203 scope
->bsm_frag_tag
= 0;
204 list_delete_all_node(scope
->bsm_list
);
206 for (rn
= route_top(scope
->bsrp_table
); rn
; rn
= route_next(rn
)) {
208 bsgrp_node
= (struct bsgrp_node
*)rn
->info
;
211 zlog_debug("%s: bsgrp_node is null", __func__
);
214 /* Give grace time for rp to continue for another hold time */
215 if ((bsgrp_node
->bsrp_list
) && (bsgrp_node
->bsrp_list
->count
)) {
216 bsrp
= listnode_head(bsgrp_node
->bsrp_list
);
217 pim_g2rp_timer_restart(bsrp
, bsrp
->rp_holdtime
);
219 /* clear pending list */
220 if ((bsgrp_node
->partial_bsrp_list
)
221 && (bsgrp_node
->partial_bsrp_list
->count
)) {
222 list_delete_all_node(bsgrp_node
->partial_bsrp_list
);
223 bsgrp_node
->pend_rp_cnt
= 0;
229 static void pim_bs_timer_stop(struct bsm_scope
*scope
)
232 zlog_debug("%s : BS timer being stopped of sz: %d", __func__
,
234 THREAD_OFF(scope
->bs_timer
);
237 static void pim_bs_timer_start(struct bsm_scope
*scope
, int bs_timeout
)
241 zlog_debug("%s : Invalid scope(NULL).", __func__
);
244 THREAD_OFF(scope
->bs_timer
);
247 "%s : starting bs timer for scope %d with timeout %d secs",
248 __func__
, scope
->sz_id
, bs_timeout
);
249 thread_add_timer(router
->master
, pim_on_bs_timer
, scope
, bs_timeout
,
253 static inline void pim_bs_timer_restart(struct bsm_scope
*scope
, int bs_timeout
)
255 pim_bs_timer_start(scope
, bs_timeout
);
258 void pim_bsm_proc_init(struct pim_instance
*pim
)
260 memset(&pim
->global_scope
, 0, sizeof(struct bsm_scope
));
262 pim
->global_scope
.sz_id
= PIM_GBL_SZ_ID
;
263 pim
->global_scope
.bsrp_table
= route_table_init();
264 pim
->global_scope
.accept_nofwd_bsm
= true;
265 pim
->global_scope
.state
= NO_INFO
;
266 pim
->global_scope
.pim
= pim
;
267 pim
->global_scope
.bsm_list
= list_new();
268 pim
->global_scope
.bsm_list
->del
= (void (*)(void *))pim_bsm_node_free
;
269 pim_bs_timer_start(&pim
->global_scope
, PIM_BS_TIME
);
272 void pim_bsm_proc_free(struct pim_instance
*pim
)
274 struct route_node
*rn
;
275 struct bsgrp_node
*bsgrp
;
277 pim_bs_timer_stop(&pim
->global_scope
);
279 if (pim
->global_scope
.bsm_list
)
280 list_delete(&pim
->global_scope
.bsm_list
);
282 for (rn
= route_top(pim
->global_scope
.bsrp_table
); rn
;
283 rn
= route_next(rn
)) {
287 pim_free_bsgrp_data(bsgrp
);
290 route_table_finish(pim
->global_scope
.bsrp_table
);
293 static bool is_hold_time_elapsed(void *data
)
295 struct bsm_rpinfo
*bsrp
;
299 if (bsrp
->elapse_time
< bsrp
->rp_holdtime
)
305 static int pim_on_g2rp_timer(struct thread
*t
)
307 struct bsm_rpinfo
*bsrp
;
308 struct bsm_rpinfo
*bsrp_node
;
309 struct bsgrp_node
*bsgrp_node
;
310 struct listnode
*bsrp_ln
;
311 struct pim_instance
*pim
;
312 struct rp_info
*rp_info
;
313 struct route_node
*rn
;
315 struct in_addr bsrp_addr
;
317 bsrp
= THREAD_ARG(t
);
318 THREAD_OFF(bsrp
->g2rp_timer
);
319 bsgrp_node
= bsrp
->bsgrp_node
;
321 /* elapse time is the hold time of expired node */
322 elapse
= bsrp
->rp_holdtime
;
323 bsrp_addr
= bsrp
->rp_address
;
325 /* update elapse for all bsrp nodes */
326 for (ALL_LIST_ELEMENTS_RO(bsgrp_node
->bsrp_list
, bsrp_ln
, bsrp_node
))
327 bsrp_node
->elapse_time
+= elapse
;
329 /* remove the expired nodes from the list */
330 list_filter_out_nodes(bsgrp_node
->bsrp_list
, is_hold_time_elapsed
);
332 /* Get the next elected rp node */
333 bsrp
= listnode_head(bsgrp_node
->bsrp_list
);
334 pim
= bsgrp_node
->scope
->pim
;
335 rn
= route_node_lookup(pim
->rp_table
, &bsgrp_node
->group
);
338 zlog_warn("%s: Route node doesn't exist", __func__
);
342 rp_info
= (struct rp_info
*)rn
->info
;
345 route_unlock_node(rn
);
349 if (rp_info
->rp_src
!= RP_SRC_STATIC
) {
350 /* If new rp available, change it else delete the existing */
352 bsrp_addr
= bsrp
->rp_address
;
353 pim_g2rp_timer_start(
354 bsrp
, (bsrp
->rp_holdtime
- bsrp
->elapse_time
));
355 pim_rp_change(pim
, bsrp_addr
, bsgrp_node
->group
,
358 pim_rp_del(pim
, bsrp_addr
, bsgrp_node
->group
, NULL
,
363 if ((!bsgrp_node
->bsrp_list
->count
)
364 && (!bsgrp_node
->partial_bsrp_list
->count
)) {
365 pim_free_bsgrp_node(pim
->global_scope
.bsrp_table
,
367 pim_free_bsgrp_data(bsgrp_node
);
373 static void pim_g2rp_timer_start(struct bsm_rpinfo
*bsrp
, int hold_time
)
377 zlog_debug("%s : Invalid brsp(NULL).", __func__
);
380 THREAD_OFF(bsrp
->g2rp_timer
);
385 "%s : starting g2rp timer for grp: %s - rp: %s with timeout %d secs(Actual Hold time : %d secs)",
386 __func__
, prefix2str(&bsrp
->bsgrp_node
->group
, buf
, 48),
387 inet_ntoa(bsrp
->rp_address
), hold_time
,
391 thread_add_timer(router
->master
, pim_on_g2rp_timer
, bsrp
, hold_time
,
395 static inline void pim_g2rp_timer_restart(struct bsm_rpinfo
*bsrp
,
398 pim_g2rp_timer_start(bsrp
, hold_time
);
401 static void pim_g2rp_timer_stop(struct bsm_rpinfo
*bsrp
)
409 zlog_debug("%s : stopping g2rp timer for grp: %s - rp: %s",
411 prefix2str(&bsrp
->bsgrp_node
->group
, buf
, 48),
412 inet_ntoa(bsrp
->rp_address
));
415 THREAD_OFF(bsrp
->g2rp_timer
);
418 static bool is_hold_time_zero(void *data
)
420 struct bsm_rpinfo
*bsrp
;
424 if (bsrp
->rp_holdtime
)
430 static void pim_instate_pend_list(struct bsgrp_node
*bsgrp_node
)
432 struct bsm_rpinfo
*active
;
433 struct bsm_rpinfo
*pend
;
435 struct rp_info
*rp_info
;
436 struct route_node
*rn
;
437 struct pim_instance
*pim
;
438 struct rp_info
*rp_all
;
439 struct prefix group_all
;
440 bool had_rp_node
= true;
442 pim
= bsgrp_node
->scope
->pim
;
443 active
= listnode_head(bsgrp_node
->bsrp_list
);
445 /* Remove nodes with hold time 0 & check if list still has a head */
446 list_filter_out_nodes(bsgrp_node
->partial_bsrp_list
, is_hold_time_zero
);
447 pend
= listnode_head(bsgrp_node
->partial_bsrp_list
);
449 if (!str2prefix("224.0.0.0/4", &group_all
))
452 rp_all
= pim_rp_find_match_group(pim
, &group_all
);
453 rn
= route_node_lookup(pim
->rp_table
, &bsgrp_node
->group
);
456 pim_g2rp_timer_start(pend
, pend
->rp_holdtime
);
458 /* if rp node doesn't exist or exist but not configured(rp_all),
459 * install the rp from head(if exists) of partial list. List is
460 * is sorted such that head is the elected RP for the group.
462 if (!rn
|| (prefix_same(&rp_all
->group
, &bsgrp_node
->group
)
463 && pim_rpf_addr_is_inaddr_none(&rp_all
->rp
))) {
465 zlog_debug("%s: Route node doesn't exist", __func__
);
467 pim_rp_new(pim
, pend
->rp_address
, bsgrp_node
->group
,
471 rp_info
= (struct rp_info
*)rn
->info
;
473 route_unlock_node(rn
);
475 pim_rp_new(pim
, pend
->rp_address
,
476 bsgrp_node
->group
, NULL
, RP_SRC_BSR
);
481 /* We didn't have rp node and pending list is empty(unlikely), cleanup*/
482 if ((!had_rp_node
) && (!pend
)) {
483 pim_free_bsgrp_node(bsgrp_node
->scope
->bsrp_table
,
485 pim_free_bsgrp_data(bsgrp_node
);
489 if ((had_rp_node
) && (rp_info
->rp_src
!= RP_SRC_STATIC
)) {
490 /* This means we searched and got rp node, needs unlock */
491 route_unlock_node(rn
);
493 if (active
&& pend
) {
494 if ((active
->rp_address
.s_addr
495 != pend
->rp_address
.s_addr
))
496 pim_rp_change(pim
, pend
->rp_address
,
497 bsgrp_node
->group
, RP_SRC_BSR
);
500 /* Possible when the first BSM has group with 0 rp count */
501 if ((!active
) && (!pend
)) {
504 "%s: Both bsrp and partial list are empty",
507 pim_free_bsgrp_node(bsgrp_node
->scope
->bsrp_table
,
509 pim_free_bsgrp_data(bsgrp_node
);
513 /* Possible when a group with 0 rp count received in BSM */
514 if ((active
) && (!pend
)) {
515 pim_rp_del(pim
, active
->rp_address
, bsgrp_node
->group
,
517 pim_free_bsgrp_node(bsgrp_node
->scope
->bsrp_table
,
520 zlog_debug("%s:Pend List is null,del grp node",
523 pim_free_bsgrp_data(bsgrp_node
);
528 if ((had_rp_node
) && (rp_info
->rp_src
== RP_SRC_STATIC
)) {
529 /* We need to unlock rn this case */
530 route_unlock_node(rn
);
531 /* there is a chance that static rp exist and bsrp cleaned
532 * so clean bsgrp node if pending list empty
537 "%s: Partial list is empty, static rp exists",
539 pim_free_bsgrp_node(bsgrp_node
->scope
->bsrp_table
,
541 pim_free_bsgrp_data(bsgrp_node
);
546 /* swap the list & delete all nodes in partial list (old bsrp_list)
548 * active is head of bsrp list
549 * pend is head of partial list
551 * active is head of partial list
552 * pend is head of bsrp list
553 * So check appriate head after swap and clean the new partial list
555 temp
= bsgrp_node
->bsrp_list
;
556 bsgrp_node
->bsrp_list
= bsgrp_node
->partial_bsrp_list
;
557 bsgrp_node
->partial_bsrp_list
= temp
;
560 pim_g2rp_timer_stop(active
);
561 list_delete_all_node(bsgrp_node
->partial_bsrp_list
);
565 static bool pim_bsr_rpf_check(struct pim_instance
*pim
, struct in_addr bsr
,
566 struct in_addr ip_src_addr
)
568 struct pim_nexthop nexthop
;
571 memset(&nexthop
, 0, sizeof(nexthop
));
573 /* New BSR recived */
574 if (bsr
.s_addr
!= pim
->global_scope
.current_bsr
.s_addr
) {
575 result
= pim_nexthop_match(pim
, bsr
, ip_src_addr
);
577 /* Nexthop lookup pass for the new BSR address */
582 char bsr_str
[INET_ADDRSTRLEN
];
584 pim_inet4_dump("<bsr?>", bsr
, bsr_str
, sizeof(bsr_str
));
585 zlog_debug("%s : No route to BSR address %s", __func__
,
591 return pim_nexthop_match_nht_cache(pim
, bsr
, ip_src_addr
);
594 static bool is_preferred_bsr(struct pim_instance
*pim
, struct in_addr bsr
,
597 if (bsr
.s_addr
== pim
->global_scope
.current_bsr
.s_addr
)
600 if (bsr_prio
> pim
->global_scope
.current_bsr_prio
)
603 else if (bsr_prio
== pim
->global_scope
.current_bsr_prio
) {
604 if (ntohl(bsr
.s_addr
)
605 >= ntohl(pim
->global_scope
.current_bsr
.s_addr
))
613 static void pim_bsm_update(struct pim_instance
*pim
, struct in_addr bsr
,
616 struct pim_nexthop_cache pnc
;
618 if (bsr
.s_addr
!= pim
->global_scope
.current_bsr
.s_addr
) {
620 char buf
[PREFIX2STR_BUFFER
];
621 bool is_bsr_tracking
= true;
623 /* De-register old BSR and register new BSR with Zebra NHT */
624 nht_p
.family
= AF_INET
;
625 nht_p
.prefixlen
= IPV4_MAX_BITLEN
;
627 if (pim
->global_scope
.current_bsr
.s_addr
!= INADDR_ANY
) {
628 nht_p
.u
.prefix4
= pim
->global_scope
.current_bsr
;
630 prefix2str(&nht_p
, buf
, sizeof(buf
));
632 "%s: Deregister BSR addr %s with Zebra NHT",
635 pim_delete_tracked_nexthop(pim
, &nht_p
, NULL
, NULL
,
639 nht_p
.u
.prefix4
= bsr
;
641 prefix2str(&nht_p
, buf
, sizeof(buf
));
643 "%s: NHT Register BSR addr %s with Zebra NHT",
647 memset(&pnc
, 0, sizeof(struct pim_nexthop_cache
));
648 pim_find_or_track_nexthop(pim
, &nht_p
, NULL
, NULL
,
649 is_bsr_tracking
, &pnc
);
650 pim
->global_scope
.current_bsr
= bsr
;
651 pim
->global_scope
.current_bsr_first_ts
=
652 pim_time_monotonic_sec();
653 pim
->global_scope
.state
= ACCEPT_PREFERRED
;
655 pim
->global_scope
.current_bsr_prio
= bsr_prio
;
656 pim
->global_scope
.current_bsr_last_ts
= pim_time_monotonic_sec();
659 static bool pim_bsm_send_intf(uint8_t *buf
, int len
, struct interface
*ifp
,
660 struct in_addr dst_addr
)
662 struct pim_interface
*pim_ifp
;
668 zlog_debug("%s: Pim interface not available for %s",
669 __func__
, ifp
->name
);
673 if (pim_ifp
->pim_sock_fd
== -1) {
675 zlog_debug("%s: Pim sock not available for %s",
676 __func__
, ifp
->name
);
680 if (pim_msg_send(pim_ifp
->pim_sock_fd
, pim_ifp
->primary_address
,
681 dst_addr
, buf
, len
, ifp
->name
)) {
682 zlog_warn("%s: Could not send BSM message on interface: %s",
683 __func__
, ifp
->name
);
687 pim_ifp
->pim_ifstat_bsm_tx
++;
688 pim_ifp
->pim
->bsm_sent
++;
692 static bool pim_bsm_frag_send(uint8_t *buf
, uint32_t len
, struct interface
*ifp
,
693 uint32_t pim_mtu
, struct in_addr dst_addr
,
696 struct bsmmsg_grpinfo
*grpinfo
, *curgrp
;
697 uint8_t *firstgrp_ptr
;
700 uint32_t parsed_len
= 0;
701 uint32_t this_pkt_rem
;
702 uint32_t copy_byte_count
;
703 uint32_t this_pkt_len
;
704 uint8_t total_rp_cnt
;
708 bool pak_pending
= false;
710 /* MTU passed here is PIM MTU (IP MTU less IP Hdr) */
711 if (pim_mtu
< (PIM_MIN_BSM_LEN
)) {
713 "%s: mtu(pim mtu: %d) size less than minimum bootstrap len",
717 "%s: mtu (pim mtu:%d) less than minimum bootstrap len",
722 pak_start
= XCALLOC(MTYPE_PIM_BSM_PKT_VAR_MEM
, pim_mtu
);
726 /* Fill PIM header later before sending packet to calc checksum */
727 pkt
+= PIM_MSG_HEADER_LEN
;
728 buf
+= PIM_MSG_HEADER_LEN
;
730 /* copy bsm header to new packet at offset of pim hdr */
731 memcpy(pkt
, buf
, PIM_BSM_HDR_LEN
);
732 pkt
+= PIM_BSM_HDR_LEN
;
733 buf
+= PIM_BSM_HDR_LEN
;
734 parsed_len
+= (PIM_MSG_HEADER_LEN
+ PIM_BSM_HDR_LEN
);
736 /* Store the position of first grp ptr, which can be reused for
737 * next packet to start filling group. old bsm header and pim hdr
738 * remains. So need not be filled again for next packet onwards.
742 /* we received mtu excluding IP hdr len as param
743 * now this_pkt_rem is mtu excluding
744 * PIM_BSM_HDR_LEN + PIM_MSG_HEADER_LEN
746 this_pkt_rem
= pim_mtu
- (PIM_BSM_HDR_LEN
+ PIM_MSG_HEADER_LEN
);
748 /* For each group till the packet length parsed */
749 while (parsed_len
< len
) {
750 /* pkt ---> fragment's current pointer
751 * buf ---> input buffer's current pointer
752 * mtu ---> size of the pim packet - PIM header
753 * curgrp ---> current group on the fragment
754 * grpinfo ---> current group on the input buffer
755 * this_pkt_rem ---> bytes remaing on the current fragment
756 * rp_fit_cnt ---> num of rp for current grp that
758 * total_rp_cnt ---> total rp present for the group in the buf
759 * frag_rp_cnt ---> no of rp for the group to be fit in
761 * this_rp_cnt ---> how many rp have we parsed
763 grpinfo
= (struct bsmmsg_grpinfo
*)buf
;
764 memcpy(pkt
, buf
, PIM_BSM_GRP_LEN
);
765 curgrp
= (struct bsmmsg_grpinfo
*)pkt
;
766 parsed_len
+= PIM_BSM_GRP_LEN
;
767 pkt
+= PIM_BSM_GRP_LEN
;
768 buf
+= PIM_BSM_GRP_LEN
;
769 this_pkt_rem
-= PIM_BSM_GRP_LEN
;
771 /* initialize rp count and total_rp_cnt before the rp loop */
773 total_rp_cnt
= grpinfo
->frag_rp_count
;
775 /* Loop till all RPs for the group parsed */
776 while (this_rp_cnt
< total_rp_cnt
) {
777 /* All RP from a group processed here.
778 * group is pointed by grpinfo.
779 * At this point make sure buf pointing to a RP
782 rp_fit_cnt
= this_pkt_rem
/ PIM_BSM_RP_LEN
;
784 /* calculate how many rp am i going to copy in
787 if (rp_fit_cnt
> (total_rp_cnt
- this_rp_cnt
))
788 frag_rp_cnt
= total_rp_cnt
- this_rp_cnt
;
790 frag_rp_cnt
= rp_fit_cnt
;
792 /* populate the frag rp count for the current grp */
793 curgrp
->frag_rp_count
= frag_rp_cnt
;
794 copy_byte_count
= frag_rp_cnt
* PIM_BSM_RP_LEN
;
796 /* copy all the rp that we are fitting in this
799 memcpy(pkt
, buf
, copy_byte_count
);
800 this_rp_cnt
+= frag_rp_cnt
;
801 buf
+= copy_byte_count
;
802 pkt
+= copy_byte_count
;
803 parsed_len
+= copy_byte_count
;
804 this_pkt_rem
-= copy_byte_count
;
806 /* Either we couldn't fit all rp for the group or the
809 if ((this_rp_cnt
< total_rp_cnt
)
811 < (PIM_BSM_GRP_LEN
+ PIM_BSM_RP_LEN
))) {
812 /* No space to fit in more rp, send this pkt */
813 this_pkt_len
= pim_mtu
- this_pkt_rem
;
814 pim_msg_build_header(pak_start
, this_pkt_len
,
815 PIM_MSG_TYPE_BOOTSTRAP
,
817 pim_bsm_send_intf(pak_start
, this_pkt_len
, ifp
,
820 /* Construct next fragment. Reuse old packet */
822 this_pkt_rem
= pim_mtu
- (PIM_BSM_HDR_LEN
823 + PIM_MSG_HEADER_LEN
);
825 /* If pkt can't accomodate next group + atleast
826 * one rp, we must break out of this inner loop
827 * and process next RP
829 if (total_rp_cnt
== this_rp_cnt
)
832 /* If some more RPs for the same group pending,
835 memcpy(pkt
, (uint8_t *)grpinfo
,
837 curgrp
= (struct bsmmsg_grpinfo
*)pkt
;
838 pkt
+= PIM_BSM_GRP_LEN
;
839 this_pkt_rem
-= PIM_BSM_GRP_LEN
;
842 /* We filled something but not yet sent out */
845 } /* while RP count */
846 } /*while parsed len */
848 /* Send if we have any unsent packet */
850 this_pkt_len
= pim_mtu
- this_pkt_rem
;
851 pim_msg_build_header(pak_start
, this_pkt_len
,
852 PIM_MSG_TYPE_BOOTSTRAP
, no_fwd
);
853 pim_bsm_send_intf(pak_start
, (pim_mtu
- this_pkt_rem
), ifp
,
856 XFREE(MTYPE_PIM_BSM_PKT_VAR_MEM
, pak_start
);
860 static void pim_bsm_fwd_whole_sz(struct pim_instance
*pim
, uint8_t *buf
,
861 uint32_t len
, int sz
)
863 struct interface
*ifp
;
864 struct pim_interface
*pim_ifp
;
865 struct in_addr dst_addr
;
870 /* For now only global scope zone is supported, so send on all
871 * pim interfaces in the vrf
873 dst_addr
= qpim_all_pim_routers_addr
;
874 FOR_ALL_INTERFACES (pim
->vrf
, ifp
) {
876 if ((!pim_ifp
) || (!pim_ifp
->bsm_enable
))
881 * When a Bootstrap message is forwarded, it is forwarded out
882 * of every multicast-capable interface that has PIM neighbors.
884 * So skipping pim interfaces with no neighbors.
886 if (listcount(pim_ifp
->pim_neighbor_list
) == 0)
889 pim_hello_require(ifp
);
890 pim_mtu
= ifp
->mtu
- MAX_IP_HDR_LEN
;
892 ret
= pim_bsm_frag_send(buf
, len
, ifp
, pim_mtu
,
895 zlog_debug("%s: pim_bsm_frag_send returned %s",
896 __func__
, ret
? "TRUE" : "FALSE");
898 pim_msg_build_header(buf
, len
, PIM_MSG_TYPE_BOOTSTRAP
,
900 if (!pim_bsm_send_intf(buf
, len
, ifp
, dst_addr
)) {
903 "%s: pim_bsm_send_intf returned false",
910 bool pim_bsm_new_nbr_fwd(struct pim_neighbor
*neigh
, struct interface
*ifp
)
912 struct in_addr dst_addr
;
913 struct pim_interface
*pim_ifp
;
914 struct bsm_scope
*scope
;
915 struct listnode
*bsm_ln
;
916 struct bsm_info
*bsminfo
;
917 char neigh_src_str
[INET_ADDRSTRLEN
];
923 pim_inet4_dump("<src?>", neigh
->source_addr
, neigh_src_str
,
924 sizeof(neigh_src_str
));
925 zlog_debug("%s: New neighbor %s seen on %s", __func__
,
926 neigh_src_str
, ifp
->name
);
931 /* DR only forwards BSM packet */
932 if (pim_ifp
->pim_dr_addr
.s_addr
== pim_ifp
->primary_address
.s_addr
) {
935 "%s: It is not DR, so don't forward BSM packet",
939 if (!pim_ifp
->bsm_enable
) {
941 zlog_debug("%s: BSM proc not enabled on %s", __func__
,
946 scope
= &pim_ifp
->pim
->global_scope
;
948 if (!scope
->bsm_list
->count
) {
950 zlog_debug("%s: BSM list for the scope is empty",
955 if (!pim_ifp
->ucast_bsm_accept
) {
956 dst_addr
= qpim_all_pim_routers_addr
;
958 zlog_debug("%s: Sending BSM mcast to %s", __func__
,
961 dst_addr
= neigh
->source_addr
;
963 zlog_debug("%s: Sending BSM ucast to %s", __func__
,
966 pim_mtu
= ifp
->mtu
- MAX_IP_HDR_LEN
;
967 pim_hello_require(ifp
);
969 for (ALL_LIST_ELEMENTS_RO(scope
->bsm_list
, bsm_ln
, bsminfo
)) {
970 if (pim_mtu
< bsminfo
->size
) {
971 ret
= pim_bsm_frag_send(bsminfo
->bsm
, bsminfo
->size
,
972 ifp
, pim_mtu
, dst_addr
, no_fwd
);
976 "%s: pim_bsm_frag_send failed",
980 /* Pim header needs to be constructed */
981 pim_msg_build_header(bsminfo
->bsm
, bsminfo
->size
,
982 PIM_MSG_TYPE_BOOTSTRAP
, no_fwd
);
983 ret
= pim_bsm_send_intf(bsminfo
->bsm
, bsminfo
->size
,
988 "%s: pim_bsm_frag_send failed",
996 struct bsgrp_node
*pim_bsm_get_bsgrp_node(struct bsm_scope
*scope
,
999 struct route_node
*rn
;
1000 struct bsgrp_node
*bsgrp
;
1002 rn
= route_node_lookup(scope
->bsrp_table
, grp
);
1005 zlog_debug("%s: Route node doesn't exist for the group",
1010 route_unlock_node(rn
);
1015 static uint32_t hash_calc_on_grp_rp(struct prefix group
, struct in_addr rp
,
1016 uint8_t hashmasklen
)
1022 uint32_t mask
= 0xffffffff;
1024 /* mask to be made zero if hashmasklen is 0 because mask << 32
1025 * may not give 0. hashmasklen can be 0 to 32.
1027 if (hashmasklen
== 0)
1030 /* in_addr stores ip in big endian, hence network byte order
1031 * convert to uint32 before processing hash
1033 grpaddr
= ntohl(group
.u
.prefix4
.s_addr
);
1034 /* Avoid shifting by 32 bit on a 32 bit register */
1036 grpaddr
= grpaddr
& ((mask
<< (32 - hashmasklen
)));
1038 grpaddr
= grpaddr
& mask
;
1039 rp_add
= ntohl(rp
.s_addr
);
1040 temp
= 1103515245 * ((1103515245 * (uint64_t)grpaddr
+ 12345) ^ rp_add
)
1042 hash
= temp
& (0x7fffffff);
1046 static bool pim_install_bsm_grp_rp(struct pim_instance
*pim
,
1047 struct bsgrp_node
*grpnode
,
1048 struct bsmmsg_rpinfo
*rp
)
1050 struct bsm_rpinfo
*bsm_rpinfo
;
1051 uint8_t hashMask_len
= pim
->global_scope
.hashMasklen
;
1053 /*memory allocation for bsm_rpinfo */
1054 bsm_rpinfo
= XCALLOC(MTYPE_PIM_BSRP_NODE
, sizeof(*bsm_rpinfo
));
1056 bsm_rpinfo
->rp_prio
= rp
->rp_pri
;
1057 bsm_rpinfo
->rp_holdtime
= rp
->rp_holdtime
;
1058 memcpy(&bsm_rpinfo
->rp_address
, &rp
->rpaddr
.addr
,
1059 sizeof(struct in_addr
));
1060 bsm_rpinfo
->elapse_time
= 0;
1062 /* Back pointer to the group node. */
1063 bsm_rpinfo
->bsgrp_node
= grpnode
;
1065 /* update hash for this rp node */
1066 bsm_rpinfo
->hash
= hash_calc_on_grp_rp(grpnode
->group
, rp
->rpaddr
.addr
,
1068 if (listnode_add_sort_nodup(grpnode
->partial_bsrp_list
, bsm_rpinfo
)) {
1071 "%s, bs_rpinfo node added to the partial bs_rplist.",
1077 zlog_debug("%s: list node not added", __func__
);
1079 XFREE(MTYPE_PIM_BSRP_NODE
, bsm_rpinfo
);
1083 static void pim_update_pending_rp_cnt(struct bsm_scope
*sz
,
1084 struct bsgrp_node
*bsgrp
,
1085 uint16_t bsm_frag_tag
,
1086 uint32_t total_rp_count
)
1088 if (bsgrp
->pend_rp_cnt
) {
1089 /* received bsm is different packet ,
1090 * it is not same fragment.
1092 if (bsm_frag_tag
!= bsgrp
->frag_tag
) {
1095 "%s,Received a new BSM ,so clear the pending bs_rpinfo list.",
1097 list_delete_all_node(bsgrp
->partial_bsrp_list
);
1098 bsgrp
->pend_rp_cnt
= total_rp_count
;
1101 bsgrp
->pend_rp_cnt
= total_rp_count
;
1103 bsgrp
->frag_tag
= bsm_frag_tag
;
1106 /* Parsing BSR packet and adding to partial list of corresponding bsgrp node */
1107 static bool pim_bsm_parse_install_g2rp(struct bsm_scope
*scope
, uint8_t *buf
,
1108 int buflen
, uint16_t bsm_frag_tag
)
1110 struct bsmmsg_grpinfo grpinfo
;
1111 struct bsmmsg_rpinfo rpinfo
;
1112 struct prefix group
;
1113 struct bsgrp_node
*bsgrp
= NULL
;
1114 int frag_rp_cnt
= 0;
1118 while (buflen
> offset
) {
1119 if (offset
+ (int)sizeof(struct bsmmsg_grpinfo
) > buflen
) {
1122 "%s: buflen received %d is less than the internal data structure of the packet would suggest",
1126 /* Extract Group tlv from BSM */
1127 memcpy(&grpinfo
, buf
, sizeof(struct bsmmsg_grpinfo
));
1129 if (PIM_DEBUG_BSM
) {
1130 char grp_str
[INET_ADDRSTRLEN
];
1132 pim_inet4_dump("<Group?>", grpinfo
.group
.addr
, grp_str
,
1135 "%s, Group %s Rpcount:%d Fragment-Rp-count:%d",
1136 __func__
, grp_str
, grpinfo
.rp_count
,
1137 grpinfo
.frag_rp_count
);
1140 buf
+= sizeof(struct bsmmsg_grpinfo
);
1141 offset
+= sizeof(struct bsmmsg_grpinfo
);
1143 if (grpinfo
.rp_count
== 0) {
1144 if (PIM_DEBUG_BSM
) {
1145 char grp_str
[INET_ADDRSTRLEN
];
1147 pim_inet4_dump("<Group?>", grpinfo
.group
.addr
,
1148 grp_str
, sizeof(grp_str
));
1149 zlog_debug("%s, Rp count is zero for group: %s",
1155 group
.family
= AF_INET
;
1156 if (grpinfo
.group
.mask
> IPV4_MAX_BITLEN
) {
1159 "%s, v4 prefix length specified: %d is too long",
1160 __func__
, grpinfo
.group
.mask
);
1163 group
.prefixlen
= grpinfo
.group
.mask
;
1164 group
.u
.prefix4
.s_addr
= grpinfo
.group
.addr
.s_addr
;
1166 /* Get the Group node for the BSM rp table */
1167 bsgrp
= pim_bsm_get_bsgrp_node(scope
, &group
);
1171 zlog_debug("%s, Create new BSM Group node.",
1174 /* create a new node to be added to the tree. */
1175 bsgrp
= pim_bsm_new_bsgrp_node(scope
->bsrp_table
,
1180 "%s, Failed to get the BSM group node.",
1185 bsgrp
->scope
= scope
;
1188 pim_update_pending_rp_cnt(scope
, bsgrp
, bsm_frag_tag
,
1190 frag_rp_cnt
= grpinfo
.frag_rp_count
;
1193 while (frag_rp_cnt
--) {
1194 if (offset
+ (int)sizeof(struct bsmmsg_rpinfo
)
1198 "%s, buflen received: %u is less than the internal data structure of the packet would suggest",
1203 /* Extract RP address tlv from BSM */
1204 memcpy(&rpinfo
, buf
, sizeof(struct bsmmsg_rpinfo
));
1205 rpinfo
.rp_holdtime
= ntohs(rpinfo
.rp_holdtime
);
1206 buf
+= sizeof(struct bsmmsg_rpinfo
);
1207 offset
+= sizeof(struct bsmmsg_rpinfo
);
1209 if (PIM_DEBUG_BSM
) {
1210 char rp_str
[INET_ADDRSTRLEN
];
1212 pim_inet4_dump("<Rpaddr?>", rpinfo
.rpaddr
.addr
,
1213 rp_str
, sizeof(rp_str
));
1215 "%s, Rp address - %s; pri:%d hold:%d",
1216 __func__
, rp_str
, rpinfo
.rp_pri
,
1217 rpinfo
.rp_holdtime
);
1220 /* Call Install api to update grp-rp mappings */
1221 if (pim_install_bsm_grp_rp(scope
->pim
, bsgrp
, &rpinfo
))
1225 bsgrp
->pend_rp_cnt
-= ins_count
;
1227 if (!bsgrp
->pend_rp_cnt
) {
1230 "%s, Recvd all the rps for this group, so bsrp list with penidng rp list.",
1232 /* replace the bsrp_list with pending list */
1233 pim_instate_pend_list(bsgrp
);
1239 int pim_bsm_process(struct interface
*ifp
, struct ip
*ip_hdr
, uint8_t *buf
,
1240 uint32_t buf_size
, bool no_fwd
)
1242 struct bsm_hdr
*bshdr
;
1243 int sz
= PIM_GBL_SZ_ID
;
1244 struct bsmmsg_grpinfo
*msg_grp
;
1245 struct pim_interface
*pim_ifp
= NULL
;
1246 struct bsm_info
*bsminfo
;
1247 struct pim_instance
*pim
;
1248 char bsr_str
[INET_ADDRSTRLEN
];
1250 bool empty_bsm
= false;
1252 /* BSM Packet acceptance validation */
1253 pim_ifp
= ifp
->info
;
1256 zlog_debug("%s: multicast not enabled on interface %s",
1257 __func__
, ifp
->name
);
1261 pim_ifp
->pim_ifstat_bsm_rx
++;
1265 /* Drop if bsm processing is disabled on interface */
1266 if (!pim_ifp
->bsm_enable
) {
1267 zlog_warn("%s: BSM not enabled on interface %s", __func__
,
1269 pim_ifp
->pim_ifstat_bsm_cfg_miss
++;
1274 if (buf_size
< (PIM_MSG_HEADER_LEN
+ sizeof(struct bsm_hdr
))) {
1277 "%s: received buffer length of %d which is too small to properly decode",
1278 __func__
, buf_size
);
1282 bshdr
= (struct bsm_hdr
*)(buf
+ PIM_MSG_HEADER_LEN
);
1283 pim_inet4_dump("<bsr?>", bshdr
->bsr_addr
.addr
, bsr_str
,
1285 pim
->global_scope
.hashMasklen
= bshdr
->hm_len
;
1286 frag_tag
= ntohs(bshdr
->frag_tag
);
1288 /* Identify empty BSM */
1289 if ((buf_size
- PIM_BSM_HDR_LEN
- PIM_MSG_HEADER_LEN
) < PIM_BSM_GRP_LEN
)
1293 msg_grp
= (struct bsmmsg_grpinfo
*)(buf
+ PIM_MSG_HEADER_LEN
1295 /* Currently we don't support scope zoned BSM */
1296 if (msg_grp
->group
.sz
) {
1299 "%s : Administratively scoped range BSM received",
1301 pim_ifp
->pim_ifstat_bsm_invalid_sz
++;
1307 /* Drop if bsr is not preferred bsr */
1308 if (!is_preferred_bsr(pim
, bshdr
->bsr_addr
.addr
, bshdr
->bsr_prio
)) {
1310 zlog_debug("%s : Received a non-preferred BSM",
1317 /* only accept no-forward BSM if quick refresh on startup */
1318 if ((pim
->global_scope
.accept_nofwd_bsm
)
1319 || (frag_tag
== pim
->global_scope
.bsm_frag_tag
)) {
1320 pim
->global_scope
.accept_nofwd_bsm
= false;
1324 "%s : nofwd_bsm received on %s when accpt_nofwd_bsm false",
1327 pim_ifp
->pim_ifstat_ucast_bsm_cfg_miss
++;
1332 /* Mulicast BSM received */
1333 if (ip_hdr
->ip_dst
.s_addr
== qpim_all_pim_routers_addr
.s_addr
) {
1335 if (!pim_bsr_rpf_check(pim
, bshdr
->bsr_addr
.addr
,
1339 "%s : RPF check fail for BSR address %s",
1345 } else if (if_lookup_exact_address(&ip_hdr
->ip_dst
, AF_INET
,
1347 /* Unicast BSM received - if ucast bsm not enabled on
1348 * the interface, drop it
1350 if (!pim_ifp
->ucast_bsm_accept
) {
1353 "%s : Unicast BSM not enabled on interface %s",
1354 __func__
, ifp
->name
);
1355 pim_ifp
->pim_ifstat_ucast_bsm_cfg_miss
++;
1362 zlog_debug("%s : Invalid destination address",
1370 zlog_debug("%s : Empty Pref BSM received", __func__
);
1372 /* Parse Update bsm rp table and install/uninstall rp if required */
1373 if (!pim_bsm_parse_install_g2rp(
1374 &pim_ifp
->pim
->global_scope
,
1375 (buf
+ PIM_BSM_HDR_LEN
+ PIM_MSG_HEADER_LEN
),
1376 (buf_size
- PIM_BSM_HDR_LEN
- PIM_MSG_HEADER_LEN
),
1378 if (PIM_DEBUG_BSM
) {
1379 zlog_debug("%s, Parsing BSM failed.", __func__
);
1384 /* Restart the bootstrap timer */
1385 pim_bs_timer_restart(&pim_ifp
->pim
->global_scope
,
1386 PIM_BSR_DEFAULT_TIMEOUT
);
1388 /* If new BSM received, clear the old bsm database */
1389 if (pim_ifp
->pim
->global_scope
.bsm_frag_tag
!= frag_tag
) {
1390 if (PIM_DEBUG_BSM
) {
1391 zlog_debug("%s: Current frag tag: %d Frag teg rcvd: %d",
1393 pim_ifp
->pim
->global_scope
.bsm_frag_tag
,
1396 list_delete_all_node(pim_ifp
->pim
->global_scope
.bsm_list
);
1397 pim_ifp
->pim
->global_scope
.bsm_frag_tag
= frag_tag
;
1400 /* update the scope information from bsm */
1401 pim_bsm_update(pim
, bshdr
->bsr_addr
.addr
, bshdr
->bsr_prio
);
1404 pim_bsm_fwd_whole_sz(pim_ifp
->pim
, buf
, buf_size
, sz
);
1405 bsminfo
= XCALLOC(MTYPE_PIM_BSM_INFO
, sizeof(struct bsm_info
));
1407 bsminfo
->bsm
= XCALLOC(MTYPE_PIM_BSM_PKT_VAR_MEM
, buf_size
);
1409 bsminfo
->size
= buf_size
;
1410 memcpy(bsminfo
->bsm
, buf
, buf_size
);
1411 listnode_add(pim_ifp
->pim
->global_scope
.bsm_list
, bsminfo
);