xref: /netbsd/usr.sbin/mrouted/route.c (revision bf9ec67e)
1 /*	$NetBSD: route.c,v 1.6 2000/03/13 23:22:54 soren Exp $	*/
2 
3 /*
4  * The mrouted program is covered by the license in the accompanying file
5  * named "LICENSE".  Use of the mrouted program represents acceptance of
6  * the terms and conditions listed in that file.
7  *
8  * The mrouted program is COPYRIGHT 1989 by The Board of Trustees of
9  * Leland Stanford Junior University.
10  */
11 
12 
13 #include "defs.h"
14 
15 
16 /*
17  * This define statement saves a lot of space later
18  */
19 #define RT_ADDR	(struct rtentry *)&routing_table
20 
21 /*
22  * Exported variables.
23  */
24 int routes_changed;			/* 1=>some routes have changed */
25 int delay_change_reports;		/* 1=>postpone change reports  */
26 
27 
28 /*
29  * The routing table is shared with prune.c , so must not be static.
30  */
31 struct rtentry *routing_table;		/* pointer to list of route entries */
32 
33 /*
34  * Private variables.
35  */
36 static struct rtentry *rtp;		/* pointer to a route entry         */
37 static struct rtentry *rt_end;		/* pointer to last route entry      */
38 unsigned int nroutes;			/* current number of route entries  */
39 
40 /*
41  * Private functions.
42  */
43 static int init_children_and_leaves	__P((struct rtentry *r,
44 						vifi_t parent));
45 static int find_route		__P((u_int32_t origin, u_int32_t mask));
46 static void create_route	__P((u_int32_t origin, u_int32_t mask));
47 static void discard_route	__P((struct rtentry *prev_r));
48 static int compare_rts		__P((const void *rt1, const void *rt2));
49 static int report_chunk		__P((struct rtentry *start_rt, vifi_t vifi,
50 						u_int32_t dst));
51 
52 /*
53  * Initialize the routing table and associated variables.
54  */
55 void
56 init_routes()
57 {
58     routing_table        = NULL;
59     rt_end		 = RT_ADDR;
60     nroutes		 = 0;
61     routes_changed       = FALSE;
62     delay_change_reports = FALSE;
63 }
64 
65 
66 /*
67  * Initialize the children and leaf bits for route 'r', along with the
68  * associated dominant, subordinate, and leaf timing data structures.
69  * Return TRUE if this changes the value of either the children or
70  * leaf bitmaps for 'r'.
71  */
72 static int
73 init_children_and_leaves(r, parent)
74     register struct rtentry *r;
75     register vifi_t parent;
76 {
77     register vifi_t vifi;
78     register struct uvif *v;
79     vifbitmap_t old_children, old_leaves;
80 
81     VIFM_COPY(r->rt_children, old_children);
82     VIFM_COPY(r->rt_leaves,   old_leaves  );
83 
84     VIFM_CLRALL(r->rt_children);
85     VIFM_CLRALL(r->rt_leaves);
86     r->rt_flags &= ~RTF_LEAF_TIMING;
87 
88     for (vifi = 0, v = uvifs; vifi < numvifs; ++vifi, ++v) {
89 	r->rt_dominants   [vifi] = 0;
90 	r->rt_subordinates[vifi] = 0;
91 
92 	if (vifi != parent && !(v->uv_flags & (VIFF_DOWN|VIFF_DISABLED))) {
93 	    VIFM_SET(vifi, r->rt_children);
94 	    if (v->uv_neighbors == NULL) {
95 		VIFM_SET(vifi, r->rt_leaves);
96 		r->rt_leaf_timers[vifi] = 0;
97 	    }
98 	    else {
99 		r->rt_leaf_timers[vifi] = LEAF_CONFIRMATION_TIME;
100 		r->rt_flags |= RTF_LEAF_TIMING;
101 	    }
102 	}
103 	else {
104 	    r->rt_leaf_timers[vifi] = 0;
105 	}
106     }
107 
108     return (!VIFM_SAME(r->rt_children, old_children) ||
109 	    !VIFM_SAME(r->rt_leaves,   old_leaves));
110 }
111 
112 
113 /*
114  * A new vif has come up -- update the children and leaf bitmaps in all route
115  * entries to take that into account.
116  */
117 void
118 add_vif_to_routes(vifi)
119     register vifi_t vifi;
120 {
121     register struct rtentry *r;
122     register struct uvif *v;
123 
124     v = &uvifs[vifi];
125     for (r = routing_table; r != NULL; r = r->rt_next) {
126 	if (r->rt_metric != UNREACHABLE &&
127 	    !VIFM_ISSET(vifi, r->rt_children)) {
128 	    VIFM_SET(vifi, r->rt_children);
129 	    r->rt_dominants   [vifi] = 0;
130 	    r->rt_subordinates[vifi] = 0;
131 	    if (v->uv_neighbors == NULL) {
132 		VIFM_SET(vifi, r->rt_leaves);
133 		r->rt_leaf_timers[vifi] = 0;
134 	    }
135 	    else {
136 		VIFM_CLR(vifi, r->rt_leaves);
137 		r->rt_leaf_timers[vifi] = LEAF_CONFIRMATION_TIME;
138 		r->rt_flags |= RTF_LEAF_TIMING;
139 	    }
140 	    update_table_entry(r);
141 	}
142     }
143 }
144 
145 
146 /*
147  * A vif has gone down -- expire all routes that have that vif as parent,
148  * and update the children bitmaps in all other route entries to take into
149  * account the failed vif.
150  */
151 void
152 delete_vif_from_routes(vifi)
153     register vifi_t vifi;
154 {
155     register struct rtentry *r;
156 
157     for (r = routing_table; r != NULL; r = r->rt_next) {
158 	if (r->rt_metric != UNREACHABLE) {
159 	    if (vifi == r->rt_parent) {
160 		del_table_entry(r, 0, DEL_ALL_ROUTES);
161 		r->rt_timer    = ROUTE_EXPIRE_TIME;
162 		r->rt_metric   = UNREACHABLE;
163 		r->rt_flags   |= RTF_CHANGED;
164 		routes_changed = TRUE;
165 	    }
166 	    else if (VIFM_ISSET(vifi, r->rt_children)) {
167 		VIFM_CLR(vifi, r->rt_children);
168 		VIFM_CLR(vifi, r->rt_leaves);
169 		r->rt_subordinates[vifi] = 0;
170 		r->rt_leaf_timers [vifi] = 0;
171 		update_table_entry(r);
172 	    }
173 	    else {
174 		r->rt_dominants[vifi] = 0;
175 	    }
176 	}
177     }
178 }
179 
180 
181 /*
182  * A neighbor has failed or become unreachable.  If that neighbor was
183  * considered a dominant or subordinate router in any route entries,
184  * take appropriate action.
185  */
186 void
187 delete_neighbor_from_routes(addr, vifi)
188     register u_int32_t addr;
189     register vifi_t vifi;
190 {
191     register struct rtentry *r;
192     register struct uvif *v;
193 
194     v = &uvifs[vifi];
195     for (r = routing_table; r != NULL; r = r->rt_next) {
196 	if (r->rt_metric != UNREACHABLE) {
197 	    if (r->rt_dominants[vifi] == addr) {
198 		VIFM_SET(vifi, r->rt_children);
199 		r->rt_dominants   [vifi] = 0;
200 		r->rt_subordinates[vifi] = 0;
201 		if (v->uv_neighbors == NULL) {
202 		    VIFM_SET(vifi, r->rt_leaves);
203 		    r->rt_leaf_timers[vifi] = 0;
204 		}
205 		else {
206 		    VIFM_CLR(vifi, r->rt_leaves);
207 		    r->rt_leaf_timers[vifi] = LEAF_CONFIRMATION_TIME;
208 		    r->rt_flags |= RTF_LEAF_TIMING;
209 		}
210 		update_table_entry(r);
211 	    }
212 	    else if (r->rt_subordinates[vifi] == addr) {
213 		r->rt_subordinates[vifi] = 0;
214 		if (v->uv_neighbors == NULL) {
215 		    VIFM_SET(vifi, r->rt_leaves);
216 		    update_table_entry(r);
217 		}
218 		else {
219 		    r->rt_leaf_timers[vifi] = LEAF_CONFIRMATION_TIME;
220 		    r->rt_flags |= RTF_LEAF_TIMING;
221 		}
222 	    }
223 	    else if (v->uv_neighbors == NULL &&
224 		     r->rt_leaf_timers[vifi] != 0) {
225 		VIFM_SET(vifi, r->rt_leaves);
226 		r->rt_leaf_timers[vifi] = 0;
227 		update_table_entry(r);
228 	    }
229 	}
230     }
231 }
232 
233 
234 /*
235  * Prepare for a sequence of ordered route updates by initializing a pointer
236  * to the start of the routing table.  The pointer is used to remember our
237  * position in the routing table in order to avoid searching from the
238  * beginning for each update; this relies on having the route reports in
239  * a single message be in the same order as the route entries in the routing
240  * table.
241  */
242 void
243 start_route_updates()
244 {
245     rtp = RT_ADDR;
246 }
247 
248 
249 /*
250  * Starting at the route entry following the one to which 'rtp' points,
251  * look for a route entry matching the specified origin and mask.  If a
252  * match is found, return TRUE and leave 'rtp' pointing at the found entry.
253  * If no match is found, return FALSE and leave 'rtp' pointing to the route
254  * entry preceding the point at which the new origin should be inserted.
255  * This code is optimized for the normal case in which the first entry to
256  * be examined is the matching entry.
257  */
258 static int
259 find_route(origin, mask)
260     register u_int32_t origin, mask;
261 {
262     register struct rtentry *r;
263 
264     r = rtp->rt_next;
265     while (r != NULL) {
266 	if (origin == r->rt_origin && mask == r->rt_originmask) {
267 	    rtp = r;
268 	    return (TRUE);
269 	}
270 	if (ntohl(mask) < ntohl(r->rt_originmask) ||
271 	    (mask == r->rt_originmask &&
272 	     ntohl(origin) < ntohl(r->rt_origin))) {
273 	    rtp = r;
274 	    r = r->rt_next;
275 	}
276 	else break;
277     }
278     return (FALSE);
279 }
280 
281 /*
282  * Create a new routing table entry for the specified origin and link it into
283  * the routing table.  The shared variable 'rtp' is assumed to point to the
284  * routing entry after which the new one should be inserted.  It is left
285  * pointing to the new entry.
286  *
287  * Only the origin, originmask, originwidth and flags fields are initialized
288  * in the new route entry; the caller is responsible for filling in the rest.
289  */
290 static void
291 create_route(origin, mask)
292     u_int32_t origin, mask;
293 {
294     register struct rtentry *r;
295 
296     if ((r = (struct rtentry *) malloc(sizeof(struct rtentry) +
297 				       (2 * numvifs * sizeof(u_int32_t)) +
298 				       (numvifs * sizeof(u_int)))) == NULL) {
299 	log(LOG_ERR, 0, "ran out of memory");	/* fatal */
300     }
301     r->rt_origin     = origin;
302     r->rt_originmask = mask;
303     if      (((char *)&mask)[3] != 0) r->rt_originwidth = 4;
304     else if (((char *)&mask)[2] != 0) r->rt_originwidth = 3;
305     else if (((char *)&mask)[1] != 0) r->rt_originwidth = 2;
306     else                              r->rt_originwidth = 1;
307     r->rt_flags        = 0;
308     r->rt_dominants    = (u_int32_t *)(r + 1);
309     r->rt_subordinates = (u_int32_t *)(r->rt_dominants + numvifs);
310     r->rt_leaf_timers  = (u_int *)(r->rt_subordinates + numvifs);
311     r->rt_groups       = NULL;
312 
313     r->rt_next = rtp->rt_next;
314     rtp->rt_next = r;
315     r->rt_prev = rtp;
316     if (r->rt_next != NULL)
317       (r->rt_next)->rt_prev = r;
318     else
319       rt_end = r;
320     rtp = r;
321     ++nroutes;
322 }
323 
324 
325 /*
326  * Discard the routing table entry following the one to which 'prev_r' points.
327  */
328 static void
329 discard_route(prev_r)
330     register struct rtentry *prev_r;
331 {
332     register struct rtentry *r;
333 
334     r = prev_r->rt_next;
335     prev_r->rt_next = r->rt_next;
336     if (prev_r->rt_next != NULL)
337       (prev_r->rt_next)->rt_prev = prev_r;
338     else
339       rt_end = prev_r;
340     free((char *)r);
341     --nroutes;
342 }
343 
344 
345 /*
346  * Process a route report for a single origin, creating or updating the
347  * corresponding routing table entry if necessary.  'src' is either the
348  * address of a neighboring router from which the report arrived, or zero
349  * to indicate a change of status of one of our own interfaces.
350  */
351 void
352 update_route(origin, mask, metric, src, vifi)
353     u_int32_t origin, mask;
354     u_int metric;
355     u_int32_t src;
356     vifi_t vifi;
357 {
358     register struct rtentry *r;
359     u_int adj_metric;
360 
361     /*
362      * Compute an adjusted metric, taking into account the cost of the
363      * subnet or tunnel over which the report arrived, and normalizing
364      * all unreachable/poisoned metrics into a single value.
365      */
366     if (src != 0 && (metric < 1 || metric >= 2*UNREACHABLE)) {
367 	log(LOG_WARNING, 0,
368 	    "%s reports out-of-range metric %u for origin %s",
369 	    inet_fmt(src, s1), metric, inet_fmts(origin, mask, s2));
370 	return;
371     }
372     adj_metric = metric + uvifs[vifi].uv_metric;
373     if (adj_metric > UNREACHABLE) adj_metric = UNREACHABLE;
374 
375     /*
376      * Look up the reported origin in the routing table.
377      */
378     if (!find_route(origin, mask)) {
379 	/*
380 	 * Not found.
381 	 * Don't create a new entry if the report says it's unreachable,
382 	 * or if the reported origin and mask are invalid.
383 	 */
384 	if (adj_metric == UNREACHABLE) {
385 	    return;
386 	}
387 	if (src != 0 && !inet_valid_subnet(origin, mask)) {
388 	    log(LOG_WARNING, 0,
389 		"%s reports an invalid origin (%s) and/or mask (%08x)",
390 		inet_fmt(src, s1), inet_fmt(origin, s2), ntohl(mask));
391 	    return;
392 	}
393 
394 	/*
395 	 * OK, create the new routing entry.  'rtp' will be left pointing
396 	 * to the new entry.
397 	 */
398 	create_route(origin, mask);
399 
400 	/*
401 	 * Now "steal away" any sources that belong under this route
402 	 * by deleting any cache entries they might have created
403 	 * and allowing the kernel to re-request them.
404 	 */
405 	steal_sources(rtp);
406 
407 	rtp->rt_metric = UNREACHABLE;	/* temporary; updated below */
408     }
409 
410     /*
411      * We now have a routing entry for the reported origin.  Update it?
412      */
413     r = rtp;
414     if (r->rt_metric == UNREACHABLE) {
415 	/*
416 	 * The routing entry is for a formerly-unreachable or new origin.
417 	 * If the report claims reachability, update the entry to use
418 	 * the reported route.
419 	 */
420 	if (adj_metric == UNREACHABLE)
421 	    return;
422 
423 	r->rt_parent   = vifi;
424 	init_children_and_leaves(r, vifi);
425 
426 	r->rt_gateway  = src;
427 	r->rt_timer    = 0;
428 	r->rt_metric   = adj_metric;
429 	r->rt_flags   |= RTF_CHANGED;
430 	routes_changed = TRUE;
431 	update_table_entry(r);
432     }
433     else if (src == r->rt_gateway) {
434 	/*
435 	 * The report has come either from the interface directly-connected
436 	 * to the origin subnet (src and r->rt_gateway both equal zero) or
437 	 * from the gateway we have chosen as the best first-hop gateway back
438 	 * towards the origin (src and r->rt_gateway not equal zero).  Reset
439 	 * the route timer and, if the reported metric has changed, update
440 	 * our entry accordingly.
441 	 */
442 	r->rt_timer = 0;
443 	if (adj_metric == r->rt_metric)
444 	    return;
445 
446 	if (adj_metric == UNREACHABLE) {
447 	    del_table_entry(r, 0, DEL_ALL_ROUTES);
448 	    r->rt_timer = ROUTE_EXPIRE_TIME;
449 	}
450 	else if (adj_metric < r->rt_metric) {
451 	    if (init_children_and_leaves(r, vifi)) {
452 		update_table_entry(r);
453 	    }
454 	}
455 	r->rt_metric   = adj_metric;
456 	r->rt_flags   |= RTF_CHANGED;
457 	routes_changed = TRUE;
458     }
459     else if (src == 0 ||
460 	     (r->rt_gateway != 0 &&
461 	      (adj_metric < r->rt_metric ||
462 	       (adj_metric == r->rt_metric &&
463 		(ntohl(src) < ntohl(r->rt_gateway) ||
464 		 r->rt_timer >= ROUTE_SWITCH_TIME))))) {
465 	/*
466 	 * The report is for an origin we consider reachable; the report
467 	 * comes either from one of our own interfaces or from a gateway
468 	 * other than the one we have chosen as the best first-hop gateway
469 	 * back towards the origin.  If the source of the update is one of
470 	 * our own interfaces, or if the origin is not a directly-connected
471 	 * subnet and the reported metric for that origin is better than
472 	 * what our routing entry says, update the entry to use the new
473 	 * gateway and metric.  We also switch gateways if the reported
474 	 * metric is the same as the one in the route entry and the gateway
475 	 * associated with the route entry has not been heard from recently,
476 	 * or if the metric is the same but the reporting gateway has a lower
477 	 * IP address than the gateway associated with the route entry.
478 	 * Did you get all that?
479 	 */
480 	if (r->rt_parent != vifi || adj_metric < r->rt_metric) {
481 	    /*
482 	     * XXX Why do we do this if we are just changing the metric?
483 	     */
484 	    r->rt_parent = vifi;
485 	    if (init_children_and_leaves(r, vifi)) {
486 		update_table_entry(r);
487 	    }
488 	}
489 	r->rt_gateway  = src;
490 	r->rt_timer    = 0;
491 	r->rt_metric   = adj_metric;
492 	r->rt_flags   |= RTF_CHANGED;
493 	routes_changed = TRUE;
494     }
495     else if (vifi != r->rt_parent) {
496 	/*
497 	 * The report came from a vif other than the route's parent vif.
498 	 * Update the children and leaf info, if necessary.
499 	 */
500 	if (VIFM_ISSET(vifi, r->rt_children)) {
501 	    /*
502 	     * Vif is a child vif for this route.
503 	     */
504 	    if (metric  < r->rt_metric ||
505 		(metric == r->rt_metric &&
506 		 ntohl(src) < ntohl(uvifs[vifi].uv_lcl_addr))) {
507 		/*
508 		 * Neighbor has lower metric to origin (or has same metric
509 		 * and lower IP address) -- it becomes the dominant router,
510 		 * and vif is no longer a child for me.
511 		 */
512 		VIFM_CLR(vifi, r->rt_children);
513 		VIFM_CLR(vifi, r->rt_leaves);
514 		r->rt_dominants   [vifi] = src;
515 		r->rt_subordinates[vifi] = 0;
516 		r->rt_leaf_timers [vifi] = 0;
517 		update_table_entry(r);
518 	    }
519 	    else if (metric > UNREACHABLE) {	/* "poisoned reverse" */
520 		/*
521 		 * Neighbor considers this vif to be on path to route's
522 		 * origin; if no subordinate recorded, record this neighbor
523 		 * as subordinate and clear the leaf flag.
524 		 */
525 		if (r->rt_subordinates[vifi] == 0) {
526 		    VIFM_CLR(vifi, r->rt_leaves);
527 		    r->rt_subordinates[vifi] = src;
528 		    r->rt_leaf_timers [vifi] = 0;
529 		    update_table_entry(r);
530 		}
531 	    }
532 	    else if (src == r->rt_subordinates[vifi]) {
533 		/*
534 		 * Current subordinate no longer considers this vif to be on
535 		 * path to route's origin; it is no longer a subordinate
536 		 * router, and we set the leaf confirmation timer to give
537 		 * us time to hear from other subordinates.
538 		 */
539 		r->rt_subordinates[vifi] = 0;
540 		if (uvifs[vifi].uv_neighbors == NULL ||
541 		    uvifs[vifi].uv_neighbors->al_next == NULL) {
542 		    VIFM_SET(vifi, r->rt_leaves);
543 		    update_table_entry(r);
544 		}
545 		else {
546 		    r->rt_leaf_timers [vifi] = LEAF_CONFIRMATION_TIME;
547 		    r->rt_flags |= RTF_LEAF_TIMING;
548 		}
549 	    }
550 
551 	}
552 	else if (src == r->rt_dominants[vifi] &&
553 		 (metric  > r->rt_metric ||
554 		  (metric == r->rt_metric &&
555 		   ntohl(src) > ntohl(uvifs[vifi].uv_lcl_addr)))) {
556 	    /*
557 	     * Current dominant no longer has a lower metric to origin
558 	     * (or same metric and lower IP address); we adopt the vif
559 	     * as our own child.
560 	     */
561 	    VIFM_SET(vifi, r->rt_children);
562 	    r->rt_dominants  [vifi] = 0;
563 	    if (metric > UNREACHABLE) {
564 		r->rt_subordinates[vifi] = src;
565 	    }
566 	    else if (uvifs[vifi].uv_neighbors == NULL ||
567 		     uvifs[vifi].uv_neighbors->al_next == NULL) {
568 		VIFM_SET(vifi, r->rt_leaves);
569 	    }
570 	    else {
571 		r->rt_leaf_timers[vifi] = LEAF_CONFIRMATION_TIME;
572 		r->rt_flags |= RTF_LEAF_TIMING;
573 	    }
574 	    update_table_entry(r);
575 	}
576     }
577 }
578 
579 
580 /*
581  * On every timer interrupt, advance the timer in each routing entry.
582  */
583 void
584 age_routes()
585 {
586     register struct rtentry *r;
587     register struct rtentry *prev_r;
588     register vifi_t vifi;
589 
590     for (prev_r = RT_ADDR, r = routing_table;
591 	 r != NULL;
592 	 prev_r = r, r = r->rt_next) {
593 
594 	if ((r->rt_timer += TIMER_INTERVAL) < ROUTE_EXPIRE_TIME) {
595 	    /*
596 	     * Route is still good; see if any leaf timers need to be
597 	     * advanced.
598 	     */
599 	    if (r->rt_flags & RTF_LEAF_TIMING) {
600 		r->rt_flags &= ~RTF_LEAF_TIMING;
601 		for (vifi = 0; vifi < numvifs; ++vifi) {
602 		    if (r->rt_leaf_timers[vifi] != 0) {
603 			/*
604 			 * Unlike other timers, leaf timers decrement.
605 			 */
606 			if ((r->rt_leaf_timers[vifi] -= TIMER_INTERVAL) == 0){
607 #ifdef NOTYET
608 			    /* If the vif is a physical leaf but has neighbors,
609 			     * it is not a tree leaf.  If I am a leaf, then no
610 			     * interface with neighbors is a tree leaf. */
611 			    if (!(((uvifs[vifi].uv_flags & VIFF_LEAF) ||
612 				   (vifs_with_neighbors == 1)) &&
613 				  (uvifs[vifi].uv_neighbors != NULL))) {
614 #endif
615 				VIFM_SET(vifi, r->rt_leaves);
616 				update_table_entry(r);
617 #ifdef NOTYET
618 			    }
619 #endif
620 			}
621 			else {
622 			    r->rt_flags |= RTF_LEAF_TIMING;
623 			}
624 		    }
625 		}
626 	    }
627 	}
628 	else if (r->rt_timer >= ROUTE_DISCARD_TIME) {
629 	    /*
630 	     * Time to garbage-collect the route entry.
631 	     */
632 	    del_table_entry(r, 0, DEL_ALL_ROUTES);
633 	    discard_route(prev_r);
634 	    r = prev_r;
635 	}
636 	else if (r->rt_metric != UNREACHABLE) {
637 	    /*
638 	     * Time to expire the route entry.  If the gateway is zero,
639 	     * i.e., it is a route to a directly-connected subnet, just
640 	     * set the timer back to zero; such routes expire only when
641 	     * the interface to the subnet goes down.
642 	     */
643 	    if (r->rt_gateway == 0) {
644 		r->rt_timer = 0;
645 	    }
646 	    else {
647 		del_table_entry(r, 0, DEL_ALL_ROUTES);
648 		r->rt_metric   = UNREACHABLE;
649 		r->rt_flags   |= RTF_CHANGED;
650 		routes_changed = TRUE;
651 	    }
652 	}
653     }
654 }
655 
656 
657 /*
658  * Mark all routes as unreachable.  This function is called only from
659  * hup() in preparation for informing all neighbors that we are going
660  * off the air.  For consistency, we ought also to delete all reachable
661  * route entries from the kernel, but since we are about to exit we rely
662  * on the kernel to do its own cleanup -- no point in making all those
663  * expensive kernel calls now.
664  */
665 void
666 expire_all_routes()
667 {
668     register struct rtentry *r;
669 
670     for (r = routing_table; r != NULL; r = r->rt_next) {
671 	r->rt_metric   = UNREACHABLE;
672 	r->rt_flags   |= RTF_CHANGED;
673 	routes_changed = TRUE;
674     }
675 }
676 
677 
678 /*
679  * Delete all the routes in the routing table.
680  */
681 void
682 free_all_routes()
683 {
684     register struct rtentry *r;
685 
686     r = RT_ADDR;
687 
688     while (r->rt_next)
689 	discard_route(r);
690 }
691 
692 
693 /*
694  * Process an incoming neighbor probe message.
695  */
696 void
697 accept_probe(src, dst, p, datalen, level)
698     u_int32_t src;
699     u_int32_t dst;
700     char *p;
701     int datalen;
702     u_int32_t level;
703 {
704     vifi_t vifi;
705 
706     if ((vifi = find_vif(src, dst)) == NO_VIF) {
707 	log(LOG_INFO, 0,
708     	    "ignoring probe from non-neighbor %s", inet_fmt(src, s1));
709 	return;
710     }
711 
712     update_neighbor(vifi, src, DVMRP_PROBE, p, datalen, level);
713 }
714 
715 struct newrt {
716 	u_int32_t mask;
717 	u_int32_t origin;
718 	int metric;
719 	int pad;
720 };
721 
722 static int
723 compare_rts(rt1, rt2)
724     const void *rt1;
725     const void *rt2;
726 {
727     register struct newrt *r1 = (struct newrt *)rt1;
728     register struct newrt *r2 = (struct newrt *)rt2;
729     register u_int32_t m1 = ntohl(r1->mask);
730     register u_int32_t m2 = ntohl(r2->mask);
731     register u_int32_t o1, o2;
732 
733     if (m1 > m2)
734 	return (-1);
735     if (m1 < m2)
736 	return (1);
737 
738     /* masks are equal */
739     o1 = ntohl(r1->origin);
740     o2 = ntohl(r2->origin);
741     if (o1 > o2)
742 	return (-1);
743     if (o1 < o2)
744 	return (1);
745     return (0);
746 }
747 
748 /*
749  * Process an incoming route report message.
750  */
751 void
752 accept_report(src, dst, p, datalen, level)
753     u_int32_t src, dst, level;
754     register char *p;
755     register int datalen;
756 {
757     vifi_t vifi;
758     register int width, i, nrt = 0;
759     int metric;
760     u_int32_t mask;
761     u_int32_t origin;
762     struct newrt rt[4096];
763 
764     if ((vifi = find_vif(src, dst)) == NO_VIF) {
765 	log(LOG_INFO, 0,
766     	    "ignoring route report from non-neighbor %s", inet_fmt(src, s1));
767 	return;
768     }
769 
770     if (!update_neighbor(vifi, src, DVMRP_REPORT, NULL, 0, level))
771 	return;
772 
773     if (datalen > 2*4096) {
774 	log(LOG_INFO, 0,
775     	    "ignoring oversize (%d bytes) route report from %s",
776 	    datalen, inet_fmt(src, s1));
777 	return;
778     }
779 
780     while (datalen > 0) {	/* Loop through per-mask lists. */
781 
782 	if (datalen < 3) {
783 	    log(LOG_WARNING, 0,
784 		"received truncated route report from %s",
785 		inet_fmt(src, s1));
786 	    return;
787 	}
788 	((u_char *)&mask)[0] = 0xff;            width = 1;
789 	if ((((u_char *)&mask)[1] = *p++) != 0) width = 2;
790 	if ((((u_char *)&mask)[2] = *p++) != 0) width = 3;
791 	if ((((u_char *)&mask)[3] = *p++) != 0) width = 4;
792 	if (!inet_valid_mask(ntohl(mask))) {
793 	    log(LOG_WARNING, 0,
794 		"%s reports bogus netmask 0x%08x (%s)",
795 		inet_fmt(src, s1), ntohl(mask), inet_fmt(mask, s2));
796 	    return;
797 	}
798 	datalen -= 3;
799 
800 	do {			/* Loop through (origin, metric) pairs */
801 	    if (datalen < width + 1) {
802 		log(LOG_WARNING, 0,
803 		    "received truncated route report from %s",
804 		    inet_fmt(src, s1));
805 		return;
806 	    }
807 	    origin = 0;
808 	    for (i = 0; i < width; ++i)
809 		((char *)&origin)[i] = *p++;
810 	    metric = *p++;
811 	    datalen -= width + 1;
812 	    rt[nrt].mask   = mask;
813 	    rt[nrt].origin = origin;
814 	    rt[nrt].metric = (metric & 0x7f);
815 	    ++nrt;
816 	} while (!(metric & 0x80));
817     }
818 
819     qsort((char*)rt, nrt, sizeof(rt[0]), compare_rts);
820     start_route_updates();
821     /*
822      * If the last entry is default, change mask from 0xff000000 to 0
823      */
824     if (rt[nrt-1].origin == 0)
825 	rt[nrt-1].mask = 0;
826 
827     log(LOG_DEBUG, 0, "Updating %d routes from %s to %s", nrt,
828 		inet_fmt(src, s1), inet_fmt(dst, s2));
829     for (i = 0; i < nrt; ++i) {
830 	if (i != 0 && rt[i].origin == rt[i-1].origin &&
831 		      rt[i].mask == rt[i-1].mask) {
832 	    log(LOG_WARNING, 0, "%s reports duplicate route for %s",
833 		inet_fmt(src, s1), inet_fmts(rt[i].origin, rt[i].mask, s2));
834 	    continue;
835 	}
836 	update_route(rt[i].origin, rt[i].mask, rt[i].metric,
837 		     src, vifi);
838     }
839 
840     if (routes_changed && !delay_change_reports)
841 	report_to_all_neighbors(CHANGED_ROUTES);
842 }
843 
844 
845 /*
846  * Send a route report message to destination 'dst', via virtual interface
847  * 'vifi'.  'which_routes' specifies ALL_ROUTES or CHANGED_ROUTES.
848  */
849 void
850 report(which_routes, vifi, dst)
851     int which_routes;
852     vifi_t vifi;
853     u_int32_t dst;
854 {
855     register struct rtentry *r;
856     register char *p;
857     register int i;
858     int datalen = 0;
859     int width = 0;
860     u_int32_t mask = 0;
861     u_int32_t src;
862     u_int32_t nflags;
863 
864     src = uvifs[vifi].uv_lcl_addr;
865 
866     p = send_buf + MIN_IP_HEADER_LEN + IGMP_MINLEN;
867 
868 #ifdef NOTYET
869     /* If I'm not a leaf, but the neighbor is a leaf, only advertise default */
870     if ((vifs_with_neighbors != 1) && (uvifs[vifi].uv_flags & VIFF_LEAF)) {
871       *p++ = 0;       /* 0xff000000 mask */
872       *p++ = 0;
873       *p++ = 0;
874       *p++ = 0;       /* class A net 0.0.0.0 == default */
875       *p++ = 0x81;    /*XXX metric 1, is this safe? */
876       datalen += 5;
877       send_igmp(src, dst, IGMP_DVMRP, DVMRP_REPORT,
878                 htonl(MROUTED_LEVEL), datalen);
879       return;
880     }
881 #endif
882 
883     nflags = (uvifs[vifi].uv_flags & VIFF_LEAF) ? 0 : LEAF_FLAGS;
884 
885     for (r = rt_end; r != RT_ADDR; r = r->rt_prev) {
886 
887 	if (which_routes == CHANGED_ROUTES && !(r->rt_flags & RTF_CHANGED))
888 	    continue;
889 
890 	/*
891 	 * If there is no room for this route in the current message,
892 	 * send the message and start a new one.
893 	 */
894 	if (datalen + ((r->rt_originmask == mask) ?
895 		       (width + 1) :
896 		       (r->rt_originwidth + 4)) > MAX_DVMRP_DATA_LEN) {
897 	    *(p-1) |= 0x80;
898 	    send_igmp(src, dst, IGMP_DVMRP, DVMRP_REPORT,
899 		      htonl(MROUTED_LEVEL | nflags), datalen);
900 
901 	    p = send_buf + MIN_IP_HEADER_LEN + IGMP_MINLEN;
902 	    datalen = 0;
903 	    mask = 0;
904 	}
905 
906 	if (r->rt_originmask != mask || datalen == 0) {
907 	    mask  = r->rt_originmask;
908 	    width = r->rt_originwidth;
909 	    if (datalen != 0) *(p-1) |= 0x80;
910 	    *p++ = ((char *)&mask)[1];
911 	    *p++ = ((char *)&mask)[2];
912 	    *p++ = ((char *)&mask)[3];
913 	    datalen += 3;
914 	}
915 
916 	for (i = 0; i < width; ++i)
917 	    *p++ = ((char *)&(r->rt_origin))[i];
918 
919 	*p++ = (r->rt_parent == vifi && r->rt_metric != UNREACHABLE) ?
920 	    (char)(r->rt_metric + UNREACHABLE) :  /* "poisoned reverse" */
921 		(char)(r->rt_metric);
922 
923 	datalen += width + 1;
924     }
925 
926     if (datalen != 0) {
927 	*(p-1) |= 0x80;
928 	send_igmp(src, dst, IGMP_DVMRP, DVMRP_REPORT,
929 		  htonl(MROUTED_LEVEL | nflags), datalen);
930     }
931 }
932 
933 
934 /*
935  * Send a route report message to all neighboring routers.
936  * 'which_routes' specifies ALL_ROUTES or CHANGED_ROUTES.
937  */
938 void
939 report_to_all_neighbors(which_routes)
940     int which_routes;
941 {
942     register vifi_t vifi;
943     register struct uvif *v;
944     register struct rtentry *r;
945     int routes_changed_before;
946 
947     /*
948      * Remember the state of the global routes_changed flag before
949      * generating the reports, and clear the flag.
950      */
951     routes_changed_before = routes_changed;
952     routes_changed = FALSE;
953 
954 
955     for (vifi = 0, v = uvifs; vifi < numvifs; ++vifi, ++v) {
956 	if (v->uv_neighbors != NULL) {
957 	    report(which_routes, vifi,
958 		   (v->uv_flags & VIFF_TUNNEL) ? v->uv_rmt_addr
959 		   : dvmrp_group);
960 	}
961     }
962 
963     /*
964      * If there were changed routes before we sent the reports AND
965      * if no new changes occurred while sending the reports, clear
966      * the change flags in the individual route entries.  If changes
967      * did occur while sending the reports, new reports will be
968      * generated at the next timer interrupt.
969      */
970     if (routes_changed_before && !routes_changed) {
971 	for (r = routing_table; r != NULL; r = r->rt_next) {
972 	    r->rt_flags &= ~RTF_CHANGED;
973 	}
974     }
975 
976     /*
977      * Set a flag to inhibit further reports of changed routes until the
978      * next timer interrupt.  This is to alleviate update storms.
979      */
980     delay_change_reports = TRUE;
981 }
982 
983 /*
984  * Send a route report message to destination 'dst', via virtual interface
985  * 'vifi'.  'which_routes' specifies ALL_ROUTES or CHANGED_ROUTES.
986  */
987 static int
988 report_chunk(start_rt, vifi, dst)
989     register struct rtentry *start_rt;
990     vifi_t vifi;
991     u_int32_t dst;
992 {
993     register struct rtentry *r;
994     register char *p;
995     register int i;
996     register int nrt = 0;
997     int datalen = 0;
998     int width = 0;
999     u_int32_t mask = 0;
1000     u_int32_t src;
1001     u_int32_t nflags;
1002 
1003     src = uvifs[vifi].uv_lcl_addr;
1004     p = send_buf + MIN_IP_HEADER_LEN + IGMP_MINLEN;
1005 
1006     nflags = (uvifs[vifi].uv_flags & VIFF_LEAF) ? 0 : LEAF_FLAGS;
1007 
1008     for (r = start_rt; r != RT_ADDR; r = r->rt_prev) {
1009 
1010 #ifdef NOTYET
1011 	/* Don't send poisoned routes back to parents if I am a leaf */
1012 	if ((vifs_with_neighbors == 1) && (r->rt_parent == vifi)
1013 		&& (r->rt_metric > 1)) {
1014 	    ++nrt;
1015 	    continue;
1016 	}
1017 #endif
1018 
1019 	/*
1020 	 * If there is no room for this route in the current message,
1021 	 * send it & return how many routes we sent.
1022 	 */
1023 	if (datalen + ((r->rt_originmask == mask) ?
1024 		       (width + 1) :
1025 		       (r->rt_originwidth + 4)) > MAX_DVMRP_DATA_LEN) {
1026 	    *(p-1) |= 0x80;
1027 	    send_igmp(src, dst, IGMP_DVMRP, DVMRP_REPORT,
1028 		      htonl(MROUTED_LEVEL | nflags), datalen);
1029 	    return (nrt);
1030 	}
1031 	if (r->rt_originmask != mask || datalen == 0) {
1032 	    mask  = r->rt_originmask;
1033 	    width = r->rt_originwidth;
1034 	    if (datalen != 0) *(p-1) |= 0x80;
1035 	    *p++ = ((char *)&mask)[1];
1036 	    *p++ = ((char *)&mask)[2];
1037 	    *p++ = ((char *)&mask)[3];
1038 	    datalen += 3;
1039 	}
1040 	for (i = 0; i < width; ++i)
1041 	    *p++ = ((char *)&(r->rt_origin))[i];
1042 
1043 	*p++ = (r->rt_parent == vifi && r->rt_metric != UNREACHABLE) ?
1044 	    (char)(r->rt_metric + UNREACHABLE) :  /* "poisoned reverse" */
1045 		(char)(r->rt_metric);
1046 	++nrt;
1047 	datalen += width + 1;
1048     }
1049     if (datalen != 0) {
1050 	*(p-1) |= 0x80;
1051 	send_igmp(src, dst, IGMP_DVMRP, DVMRP_REPORT,
1052 		  htonl(MROUTED_LEVEL | nflags), datalen);
1053     }
1054     return (nrt);
1055 }
1056 
1057 /*
1058  * send the next chunk of our routing table to all neighbors.
1059  * return the length of the smallest chunk we sent out.
1060  */
1061 int
1062 report_next_chunk()
1063 {
1064     register vifi_t vifi;
1065     register struct uvif *v;
1066     register struct rtentry *sr;
1067     register int i, n = 0, min = 20000;
1068     static int start_rt;
1069 
1070     if (nroutes <= 0)
1071 	return (0);
1072 
1073     /*
1074      * find this round's starting route.
1075      */
1076     for (sr = rt_end, i = start_rt; --i >= 0; ) {
1077 	sr = sr->rt_prev;
1078 	if (sr == RT_ADDR)
1079 	    sr = rt_end;
1080     }
1081 
1082     /*
1083      * send one chunk of routes starting at this round's start to
1084      * all our neighbors.
1085      */
1086     for (vifi = 0, v = uvifs; vifi < numvifs; ++vifi, ++v) {
1087 	if ((v->uv_neighbors != NULL)
1088 #ifdef NOTYET
1089 	&& !(v->uv_flags & VIFF_LEAF)
1090 #endif
1091 		) {
1092 	    n = report_chunk(sr, vifi,
1093 			     (v->uv_flags & VIFF_TUNNEL) ? v->uv_rmt_addr
1094 			     : dvmrp_group);
1095 	    if (n < min)
1096 		min = n;
1097 	}
1098     }
1099     if (min == 20000)
1100 	min = 0;	/* Neighborless router didn't send any routes */
1101 
1102     n = min;
1103     log(LOG_INFO, 0, "update %d starting at %d of %d",
1104 	n, (nroutes - start_rt), nroutes);
1105 
1106     start_rt = (start_rt + n) % nroutes;
1107     return (n);
1108 }
1109 
1110 
1111 /*
1112  * Print the contents of the routing table on file 'fp'.
1113  */
1114 void
1115 dump_routes(fp)
1116     FILE *fp;
1117 {
1118     register struct rtentry *r;
1119     register vifi_t i;
1120 
1121 
1122     fprintf(fp,
1123 	    "Multicast Routing Table (%u %s)\n%s\n",
1124 	    nroutes, (nroutes == 1) ? "entry" : "entries",
1125 	    " Origin-Subnet      From-Gateway    Metric Tmr In-Vif  Out-Vifs");
1126 
1127     for (r = routing_table; r != NULL; r = r->rt_next) {
1128 
1129 	fprintf(fp, " %-18s %-15s ",
1130 		inet_fmts(r->rt_origin, r->rt_originmask, s1),
1131 		(r->rt_gateway == 0) ? "" : inet_fmt(r->rt_gateway, s2));
1132 
1133 	fprintf(fp, (r->rt_metric == UNREACHABLE) ? "  NR " : "%4u ",
1134 		r->rt_metric);
1135 
1136 	fprintf(fp, "  %3u %3u   ", r->rt_timer, r->rt_parent);
1137 
1138 	for (i = 0; i < numvifs; ++i) {
1139 	    if (VIFM_ISSET(i, r->rt_children)) {
1140 		fprintf(fp, " %u%c",
1141 			i, VIFM_ISSET(i, r->rt_leaves) ? '*' : ' ');
1142 	    }
1143 	}
1144 	fprintf(fp, "\n");
1145     }
1146     fprintf(fp, "\n");
1147 }
1148 
1149 struct rtentry *
1150 determine_route(src)
1151     u_int32_t src;
1152 {
1153     struct rtentry *rt;
1154 
1155     for (rt = routing_table; rt != NULL; rt = rt->rt_next) {
1156 	if (rt->rt_origin == (src & rt->rt_originmask))
1157 	    break;
1158     }
1159     return rt;
1160 }
1161