Check if the nexthop neighbor for a given route has disappeared. If so, we drop the route too. If we happen to be the RPL root, we also initiate a global repair as the neighbor may moved.

core/net/tcpip.c

@@ -561,14 +561,22 @@ tcpip_ipv6_output(void)
   if(!uip_is_addr_mcast(&UIP_IP_BUF->destipaddr)) {
     /* Next hop determination */
     nbr = NULL;
+
+    /* We first check if the destination address is on our immediate
+       link. If so, we simply use the destination address as our
+       nexthop address. */
     if(uip_ds6_is_addr_onlink(&UIP_IP_BUF->destipaddr)){
       nexthop = &UIP_IP_BUF->destipaddr;
     } else {
-      uip_ds6_route_t* locrt;
+      uip_ds6_route_t *route;
-      locrt = uip_ds6_route_lookup(&UIP_IP_BUF->destipaddr);
+      /* Check if we have a route to the destination address. */
-      if(locrt == NULL) {
+      route = uip_ds6_route_lookup(&UIP_IP_BUF->destipaddr);
+
+      /* No route was found - we send to the default route instead. */
+      if(route == NULL) {
         PRINTF("tcpip_ipv6_output: no route found, using default route\n");
-        if((nexthop = uip_ds6_defrt_choose()) == NULL) {
+        nexthop = uip_ds6_defrt_choose();
+        if(nexthop == NULL) {
 #ifdef UIP_FALLBACK_INTERFACE
 	  PRINTF("FALLBACK: removing ext hdrs & setting proto %d %d\n", 
 		 uip_ext_len, *((uint8_t *)UIP_IP_BUF + 40));
@@ -586,26 +594,61 @@ tcpip_ipv6_output(void)
           uip_len = 0;
           return;
         }
+
       } else {
-        nexthop = uip_ds6_route_nexthop(locrt);
+        /* A route was found, so we look up the nexthop neighbor for
+           the route. */
+        nexthop = uip_ds6_route_nexthop(route);
+
+        /* If the nexthop is dead, for example because the neighbor
+           never responded to link-layer acks, we drop its route. */
+        if(nexthop == NULL) {
+#if UIP_CONF_IPV6_RPL
+          /* If we are running RPL, and if we are the root of the
+             network, we'll trigger a global repair berfore we remove
+             the route. */
+          rpl_dag_t *dag;
+          rpl_instance_t *instance;
+
+          dag = (rpl_dag_t *)route->state.dag;
+          if(dag != NULL) {
+            instance = dag->instance;
+
+            rpl_repair_root(instance->instance_id);
+          }
+#endif /* UIP_CONF_RPL */
+          uip_ds6_route_rm(route);
+
+          /* We don't have a nexthop to send the packet to, so we drop
+             it. */
+          return;
+        }
       }
-      if(nexthop != NULL) {
-        PRINTF("tcpip_ipv6_output: next hop ");
-        PRINT6ADDR(nexthop);
-        PRINTF("\n");
 #if TCPIP_CONF_ANNOTATE_TRANSMISSIONS
+      if(nexthop != NULL) {
+        static uint8_t annotate_last;
+        static uint8_t annotate_has_last = 0;
+
+        if(annotate_has_last) {
+          printf("#L %u 0; red\n", annotate_last);
+        }
         printf("#L %u 1; red\n", nexthop->u8[sizeof(uip_ipaddr_t) - 1]);
+        annotate_last = nexthop->u8[sizeof(uip_ipaddr_t) - 1];
+        annotate_has_last = 1;
+      }
 #endif /* TCPIP_CONF_ANNOTATE_TRANSMISSIONS */
     }
-    }
+
     /* End of next hop determination */
+
 #if UIP_CONF_IPV6_RPL
     if(rpl_update_header_final(nexthop)) {
       uip_len = 0;
       return;
     }
 #endif /* UIP_CONF_IPV6_RPL */
-    if((nbr = uip_ds6_nbr_lookup(nexthop)) == NULL) {
+    nbr = uip_ds6_nbr_lookup(nexthop);
+    if(nbr == NULL) {
 #if UIP_ND6_SEND_NA
       if((nbr = uip_ds6_nbr_add(nexthop, NULL, 0, NBR_INCOMPLETE)) == NULL) {
         uip_len = 0;
@@ -679,8 +722,8 @@ tcpip_ipv6_output(void)
       uip_len = 0;
       return;
     }
+    return;
   }
-
   /* Multicast IP destination address. */
   tcpip_output(NULL);
   uip_len = 0;