1 files changed, 85 insertions, 51 deletions

diff --git a/zebra/zebra_rib.c b/zebra/zebra_rib.c
index 0e1df1cc35..b31b6a1250 100644
--- a/zebra/zebra_rib.c
+++ b/zebra/zebra_rib.c
@@ -71,7 +71,7 @@ extern int allow_delete;
 /* Each route type's string and default distance value. */
 static const struct {
 	int key;
-	int distance;
+	uint8_t distance;
 	uint8_t meta_q_map;
 } route_info[ZEBRA_ROUTE_MAX] = {
 	[ZEBRA_ROUTE_SYSTEM] = {ZEBRA_ROUTE_SYSTEM, 0, 4},
@@ -98,6 +98,11 @@ static const struct {
 	[ZEBRA_ROUTE_BGP_DIRECT_EXT] = {ZEBRA_ROUTE_BGP_DIRECT_EXT, 20, 3},
 	[ZEBRA_ROUTE_BABEL] = {ZEBRA_ROUTE_BABEL, 100, 2},
 	[ZEBRA_ROUTE_SHARP] = {ZEBRA_ROUTE_SHARP, 150, 4},
+	[ZEBRA_ROUTE_PBR] = {ZEBRA_ROUTE_PBR, 200, 4},
+	[ZEBRA_ROUTE_BFD] = {ZEBRA_ROUTE_BFD, 255, 4},
+	[ZEBRA_ROUTE_OPENFABRIC] = {ZEBRA_ROUTE_OPENFABRIC, 115, 2},
+	[ZEBRA_ROUTE_VRRP] = {ZEBRA_ROUTE_VRRP, 255, 4}
+	/* Any new route type added to zebra, should be mirrored here */
 
 	/* no entry/default: 150 */
 };
@@ -161,8 +166,7 @@ int is_zebra_valid_kernel_table(uint32_t table_id)
 
 int is_zebra_main_routing_table(uint32_t table_id)
 {
-	if ((table_id == RT_TABLE_MAIN)
-	    || (table_id == zrouter.rtm_table_default))
+	if (table_id == RT_TABLE_MAIN)
 		return 1;
 	return 0;
 }
@@ -470,8 +474,7 @@ static int nexthop_active(afi_t afi, struct route_entry *re,
 			if (IS_ZEBRA_DEBUG_RIB_DETAILED)
 				zlog_debug(
 					"\t%s: Interface %s is not unnumbered",
-					__PRETTY_FUNCTION__,
-					ifp ? ifp->name : "Unknown");
+					__PRETTY_FUNCTION__, ifp->name);
 			return 0;
 		}
 	}
@@ -795,12 +798,6 @@ struct route_entry *rib_lookup_ipv4(struct prefix_ipv4 *p, vrf_id_t vrf_id)
 	return NULL;
 }
 
-#define RIB_SYSTEM_ROUTE(R)                                                    \
-	((R)->type == ZEBRA_ROUTE_KERNEL || (R)->type == ZEBRA_ROUTE_CONNECT)
-
-#define RIB_KERNEL_ROUTE(R)						\
-	((R)->type == ZEBRA_ROUTE_KERNEL)
-
 /* This function verifies reachability of one given nexthop, which can be
  * numbered or unnumbered, IPv4 or IPv6. The result is unconditionally stored
  * in nexthop->flags field. The nexthop->ifindex will be updated
@@ -959,7 +956,8 @@ static int nexthop_active_update(struct route_node *rn, struct route_entry *re)
 		 * decision point.
 		 */
 		new_active = nexthop_active_check(rn, re, nexthop);
-		if (new_active && re->nexthop_active_num >= multipath_num) {
+		if (new_active
+		    && re->nexthop_active_num >= zrouter.multipath_num) {
 			UNSET_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE);
 			new_active = 0;
 		}
@@ -1172,7 +1170,7 @@ static void rib_uninstall(struct route_node *rn, struct route_entry *re)
  */
 static int rib_can_delete_dest(rib_dest_t *dest)
 {
-	if (dest->routes) {
+	if (re_list_first(&dest->routes)) {
 		return 0;
 	}
 
@@ -1200,7 +1198,6 @@ static int rib_can_delete_dest(rib_dest_t *dest)
 void zebra_rib_evaluate_rn_nexthops(struct route_node *rn, uint32_t seq)
 {
 	rib_dest_t *dest = rib_dest_from_rnode(rn);
-	struct listnode *node, *nnode;
 	struct rnh *rnh;
 
 	/*
@@ -1232,7 +1229,7 @@ void zebra_rib_evaluate_rn_nexthops(struct route_node *rn, uint32_t seq)
 		 * nht resolution and as such we need to call the
 		 * nexthop tracking evaluation code
 		 */
-		for (ALL_LIST_ELEMENTS(dest->nht, node, nnode, rnh)) {
+		frr_each (rnh_list, &dest->nht, rnh) {
 			struct zebra_vrf *zvrf =
 				zebra_vrf_lookup_by_id(rnh->vrf_id);
 			struct prefix *p = &rnh->node->p;
@@ -1308,7 +1305,7 @@ int rib_gc_dest(struct route_node *rn)
 	zebra_rib_evaluate_rn_nexthops(rn, zebra_router_get_next_sequence());
 
 	dest->rnode = NULL;
-	list_delete(&dest->nht);
+	rnh_list_fini(&dest->nht);
 	XFREE(MTYPE_RIB_DEST, dest);
 	rn->info = NULL;
 
@@ -2353,7 +2350,7 @@ rib_dest_t *zebra_rib_create_dest(struct route_node *rn)
 	rib_dest_t *dest;
 
 	dest = XCALLOC(MTYPE_RIB_DEST, sizeof(rib_dest_t));
-	dest->nht = list_new();
+	rnh_list_init(&dest->nht);
 	route_lock_node(rn); /* rn route table reference */
 	rn->info = dest;
 	dest->rnode = rn;
@@ -2401,7 +2398,6 @@ rib_dest_t *zebra_rib_create_dest(struct route_node *rn)
 /* Add RE to head of the route node. */
 static void rib_link(struct route_node *rn, struct route_entry *re, int process)
 {
-	struct route_entry *head;
 	rib_dest_t *dest;
 	afi_t afi;
 	const char *rmap_name;
@@ -2416,12 +2412,7 @@ static void rib_link(struct route_node *rn, struct route_entry *re, int process)
 		dest = zebra_rib_create_dest(rn);
 	}
 
-	head = dest->routes;
-	if (head) {
-		head->prev = re;
-	}
-	re->next = head;
-	dest->routes = re;
+	re_list_add_head(&dest->routes, re);
 
 	afi = (rn->p.family == AF_INET)
 		      ? AFI_IP
@@ -2471,14 +2462,7 @@ void rib_unlink(struct route_node *rn, struct route_entry *re)
 
 	dest = rib_dest_from_rnode(rn);
 
-	if (re->next)
-		re->next->prev = re->prev;
-
-	if (re->prev)
-		re->prev->next = re->next;
-	else {
-		dest->routes = re->next;
-	}
+	re_list_del(&dest->routes, re);
 
 	if (dest->selected_fib == re)
 		dest->selected_fib = NULL;
@@ -2655,7 +2639,6 @@ void rib_lookup_and_pushup(struct prefix_ipv4 *p, vrf_id_t vrf_id)
 {
 	struct route_table *table;
 	struct route_node *rn;
-	unsigned changed = 0;
 	rib_dest_t *dest;
 
 	if (NULL == (table = zebra_vrf_table(AFI_IP, SAFI_UNICAST, vrf_id))) {
@@ -2682,7 +2665,6 @@ void rib_lookup_and_pushup(struct prefix_ipv4 *p, vrf_id_t vrf_id)
 	 * of the rest of the RE.
 	 */
 	if (dest->selected_fib) {
-		changed = 1;
 		if (IS_ZEBRA_DEBUG_RIB) {
 			char buf[PREFIX_STRLEN];
 
@@ -2692,9 +2674,8 @@ void rib_lookup_and_pushup(struct prefix_ipv4 *p, vrf_id_t vrf_id)
 			route_entry_dump(&rn->p, NULL, dest->selected_fib);
 		}
 		rib_uninstall(rn, dest->selected_fib);
-	}
-	if (changed)
 		rib_queue_add(rn);
+	}
 }
 
 int rib_add_multipath(afi_t afi, safi_t safi, struct prefix *p,
@@ -2874,7 +2855,11 @@ void rib_delete(afi_t afi, safi_t safi, vrf_id_t vrf_id, int type,
 				break;
 			}
 			for (ALL_NEXTHOPS(re->ng, rtnh))
-				if (nexthop_same_no_recurse(rtnh, nh)) {
+				/*
+				 * No guarantee all kernel send nh with labels
+				 * on delete.
+				 */
+				if (nexthop_same_no_labels(rtnh, nh)) {
 					same = re;
 					break;
 				}
@@ -3018,7 +3003,7 @@ int rib_add(afi_t afi, safi_t safi, vrf_id_t vrf_id, int type,
 	re->table = table_id;
 	re->vrf_id = vrf_id;
 	re->nexthop_num = 0;
-	re->uptime = time(NULL);
+	re->uptime = monotime(NULL);
 	re->tag = tag;
 
 	/* Add nexthop. */
@@ -3069,6 +3054,8 @@ void rib_update_table(struct route_table *table, rib_update_event_t event)
 					continue;
 
 				if (re->type != ZEBRA_ROUTE_STATIC) {
+					SET_FLAG(re->status,
+						 ROUTE_ENTRY_CHANGED);
 					rib_queue_add(rn);
 					continue;
 				}
@@ -3082,8 +3069,11 @@ void rib_update_table(struct route_table *table, rib_update_event_t event)
 				 * gateway, NHT will
 				 * take care.
 				 */
-				if (nh)
+				if (nh) {
+					SET_FLAG(re->status,
+						 ROUTE_ENTRY_CHANGED);
 					rib_queue_add(rn);
+				}
 			}
 			break;
 
@@ -3093,8 +3083,12 @@ void rib_update_table(struct route_table *table, rib_update_event_t event)
 			 * protocol in
 			 * some cases (TODO).
 			 */
-			if (rnode_to_ribs(rn))
+			if (rnode_to_ribs(rn)) {
+				RNODE_FOREACH_RE_SAFE (rn, re, next)
+					SET_FLAG(re->status,
+						 ROUTE_ENTRY_CHANGED);
 				rib_queue_add(rn);
+			}
 			break;
 
 		default:
@@ -3137,6 +3131,7 @@ void rib_sweep_table(struct route_table *table)
 
 	for (rn = route_top(table); rn; rn = srcdest_route_next(rn)) {
 		RNODE_FOREACH_RE_SAFE (rn, re, next) {
+
 			if (IS_ZEBRA_DEBUG_RIB)
 				route_entry_dump(&rn->p, NULL, re);
 
@@ -3147,6 +3142,14 @@ void rib_sweep_table(struct route_table *table)
 				continue;
 
 			/*
+			 * If routes are older than startup_time then
+			 * we know we read them in from the kernel.
+			 * As such we can safely remove them.
+			 */
+			if (zrouter.startup_time < re->uptime)
+				continue;
+
+			/*
 			 * So we are starting up and have received
 			 * routes from the kernel that we have installed
 			 * from a previous run of zebra but not cleaned
@@ -3175,7 +3178,7 @@ void rib_sweep_table(struct route_table *table)
 }
 
 /* Sweep all RIB tables.  */
-void rib_sweep_route(void)
+int rib_sweep_route(struct thread *t)
 {
 	struct vrf *vrf;
 	struct zebra_vrf *zvrf;
@@ -3189,6 +3192,8 @@ void rib_sweep_route(void)
 	}
 
 	zebra_router_sweep_route();
+
+	return 0;
 }
 
 /* Remove specific by protocol routes from 'table'. */
@@ -3219,18 +3224,23 @@ unsigned long rib_score_proto(uint8_t proto, unsigned short instance)
 {
 	struct vrf *vrf;
 	struct zebra_vrf *zvrf;
+	struct other_route_table *ort;
 	unsigned long cnt = 0;
 
-	RB_FOREACH (vrf, vrf_id_head, &vrfs_by_id)
-		if ((zvrf = vrf->info) != NULL)
-			cnt += rib_score_proto_table(
-				       proto, instance,
-				       zvrf->table[AFI_IP][SAFI_UNICAST])
-			       + rib_score_proto_table(
-					 proto, instance,
-					 zvrf->table[AFI_IP6][SAFI_UNICAST]);
+	RB_FOREACH (vrf, vrf_id_head, &vrfs_by_id) {
+		zvrf = vrf->info;
+		if (!zvrf)
+			continue;
+
+		cnt += rib_score_proto_table(proto, instance,
+					     zvrf->table[AFI_IP][SAFI_UNICAST])
+		       + rib_score_proto_table(
+			       proto, instance,
+			       zvrf->table[AFI_IP6][SAFI_UNICAST]);
 
-	cnt += zebra_router_score_proto(proto, instance);
+		frr_each(otable, &zvrf->other_tables, ort) cnt +=
+			rib_score_proto_table(proto, instance, ort->table);
+	}
 
 	return cnt;
 }
@@ -3380,9 +3390,33 @@ static int rib_dplane_results(struct dplane_ctx_q *ctxlist)
 	return 0;
 }
 
+/*
+ * Ensure there are no empty slots in the route_info array.
+ * Every route type in zebra should be present there.
+ */
+static void check_route_info(void)
+{
+	int len = array_size(route_info);
+
+	/*
+	 * ZEBRA_ROUTE_SYSTEM is special cased since
+	 * its key is 0 anyway.
+	 *
+	 * ZEBRA_ROUTE_ALL is also ignored.
+	 */
+	for (int i = 0; i < len; i++) {
+		if (i == ZEBRA_ROUTE_SYSTEM || i == ZEBRA_ROUTE_ALL)
+			continue;
+		assert(route_info[i].key);
+		assert(route_info[i].meta_q_map < MQ_SIZE);
+	}
+}
+
 /* Routing information base initialize. */
 void rib_init(void)
 {
+	check_route_info();
+
 	rib_queue_init();
 
 	/* Init dataplane, and register for results */
@@ -3451,7 +3485,7 @@ struct route_table *rib_tables_iter_next(rib_tables_iter_t *iter)
 		while (1) {
 
 			while (iter->afi_safi_ix
-			       < (int)ZEBRA_NUM_OF(afi_safis)) {
+			       < (int)array_size(afi_safis)) {
 				table = zebra_vrf_table(
 					afi_safis[iter->afi_safi_ix].afi,
 					afi_safis[iter->afi_safi_ix].safi,