Beispiel #1
0
static void dump_lid_matrix(cl_map_item_t * item, FILE * file, void *cxt)
{
	osm_switch_t *p_sw = (osm_switch_t *) item;
	osm_opensm_t *p_osm = cxt;
	osm_node_t *p_node = p_sw->p_node;
	unsigned max_lid = p_sw->max_lid_ho;
	unsigned max_port = p_sw->num_ports;
	uint16_t lid;
	uint8_t port;

	fprintf(file, "Switch: guid 0x%016" PRIx64 "\n",
		cl_ntoh64(osm_node_get_node_guid(p_node)));
	for (lid = 1; lid <= max_lid; lid++) {
		osm_port_t *p_port;
		if (osm_switch_get_least_hops(p_sw, lid) == OSM_NO_PATH)
			continue;
		fprintf(file, "0x%04x:", lid);
		for (port = 0; port < max_port; port++)
			fprintf(file, " %02x",
				osm_switch_get_hop_count(p_sw, lid, port));
		p_port = osm_get_port_by_lid_ho(&p_osm->subn, lid);
		if (p_port)
			fprintf(file, " # portguid 0x%016" PRIx64,
				cl_ntoh64(osm_port_get_guid(p_port)));
		fprintf(file, "\n");
	}
}
static void ucast_mgr_process_neighbor(IN osm_ucast_mgr_t * p_mgr,
				       IN osm_switch_t * p_this_sw,
				       IN osm_switch_t * p_remote_sw,
				       IN uint8_t port_num,
				       IN uint8_t remote_port_num)
{
	osm_switch_t *p_sw;
	cl_map_item_t *item;
	uint16_t lid_ho;
	uint16_t hops;
	osm_physp_t *p;

	OSM_LOG_ENTER(p_mgr->p_log);

	OSM_LOG(p_mgr->p_log, OSM_LOG_DEBUG,
		"Node 0x%" PRIx64 ", remote node 0x%" PRIx64
		", port %u, remote port %u\n",
		cl_ntoh64(osm_node_get_node_guid(p_this_sw->p_node)),
		cl_ntoh64(osm_node_get_node_guid(p_remote_sw->p_node)),
		port_num, remote_port_num);

	p = osm_node_get_physp_ptr(p_this_sw->p_node, port_num);

	for (item = cl_qmap_head(&p_mgr->p_subn->sw_guid_tbl);
	     item != cl_qmap_end(&p_mgr->p_subn->sw_guid_tbl);
	     item = cl_qmap_next(item)) {
		p_sw = (osm_switch_t *) item;
		lid_ho = cl_ntoh16(osm_node_get_base_lid(p_sw->p_node, 0));
		hops = osm_switch_get_least_hops(p_remote_sw, lid_ho);
		if (hops == OSM_NO_PATH)
			continue;
		hops += p->hop_wf;
		if (hops <
		    osm_switch_get_hop_count(p_this_sw, lid_ho, port_num)) {
			if (osm_switch_set_hops
			    (p_this_sw, lid_ho, port_num, (uint8_t) hops) != 0)
				OSM_LOG(p_mgr->p_log, OSM_LOG_ERROR, "ERR 3A03: "
					"cannot set hops for lid %u at switch 0x%"
					PRIx64 "\n", lid_ho,
					cl_ntoh64(osm_node_get_node_guid
						  (p_this_sw->p_node)));
			p_mgr->some_hop_count_set = TRUE;
		}
	}

	OSM_LOG_EXIT(p_mgr->p_log);
}
Beispiel #3
0
static void
__osm_ucast_mgr_process_neighbor(IN osm_ucast_mgr_t * const p_mgr,
				 IN osm_switch_t * const p_this_sw,
				 IN osm_switch_t * const p_remote_sw,
				 IN const uint8_t port_num,
				 IN const uint8_t remote_port_num)
{
	osm_switch_t *p_sw, *p_next_sw;
	uint16_t lid_ho;
	uint8_t hops;

	OSM_LOG_ENTER(p_mgr->p_log);

	OSM_LOG(p_mgr->p_log, OSM_LOG_DEBUG,
		"Node 0x%" PRIx64 ", remote node 0x%" PRIx64
		", port %u, remote port %u\n",
		cl_ntoh64(osm_node_get_node_guid(p_this_sw->p_node)),
		cl_ntoh64(osm_node_get_node_guid(p_remote_sw->p_node)),
		port_num, remote_port_num);

	p_next_sw = (osm_switch_t *) cl_qmap_head(&p_mgr->p_subn->sw_guid_tbl);
	while (p_next_sw !=
	       (osm_switch_t *) cl_qmap_end(&p_mgr->p_subn->sw_guid_tbl)) {
		p_sw = p_next_sw;
		p_next_sw = (osm_switch_t *) cl_qmap_next(&p_sw->map_item);
		lid_ho = osm_node_get_base_lid(p_sw->p_node, 0);
		lid_ho = cl_ntoh16(lid_ho);
		hops = osm_switch_get_least_hops(p_remote_sw, lid_ho);
		if (hops == OSM_NO_PATH)
			continue;
		hops++;
		if (hops <
		    osm_switch_get_hop_count(p_this_sw, lid_ho, port_num)) {
			if (osm_switch_set_hops
			    (p_this_sw, lid_ho, port_num, hops) != 0)
				OSM_LOG(p_mgr->p_log, OSM_LOG_ERROR,
					"cannot set hops for lid %u at switch 0x%"
					PRIx64 "\n", lid_ho,
					cl_ntoh64(osm_node_get_node_guid
						  (p_this_sw->p_node)));
			p_mgr->some_hop_count_set = TRUE;
		}
	}

	OSM_LOG_EXIT(p_mgr->p_log);
}
uint8_t osm_switch_recommend_mcast_path(IN osm_switch_t * p_sw,
					IN osm_port_t * p_port,
					IN uint16_t mlid_ho,
					IN boolean_t ignore_existing)
{
	uint16_t base_lid;
	uint8_t hops;
	uint8_t port_num;
	uint8_t num_ports;
	uint8_t least_hops;

	CL_ASSERT(mlid_ho >= IB_LID_MCAST_START_HO);

	if (p_port->p_node->sw) {
		if (p_port->p_node->sw == p_sw)
			return 0;
		base_lid = osm_port_get_base_lid(p_port);
	} else {
		osm_physp_t *p_physp = p_port->p_physp;
		if (!p_physp || !p_physp->p_remote_physp ||
		    !p_physp->p_remote_physp->p_node->sw)
			return OSM_NO_PATH;
		if (p_physp->p_remote_physp->p_node->sw == p_sw)
			return p_physp->p_remote_physp->port_num;
		base_lid =
		    osm_node_get_base_lid(p_physp->p_remote_physp->p_node, 0);
	}
	base_lid = cl_ntoh16(base_lid);
	num_ports = p_sw->num_ports;

	/*
	   If the user wants us to ignore existing multicast routes,
	   then simply return the shortest hop count path to the
	   target port.

	   Otherwise, return the first port that has a path to the target,
	   picking from the ports that are already in the multicast group.
	 */
	if (!ignore_existing) {
		for (port_num = 1; port_num < num_ports; port_num++) {
			if (!osm_mcast_tbl_is_port
			    (&p_sw->mcast_tbl, mlid_ho, port_num))
				continue;
			/*
			   Don't be too trusting of the current forwarding table!
			   Verify that the LID is reachable through this port.
			 */
			hops =
			    osm_switch_get_hop_count(p_sw, base_lid, port_num);
			if (hops != OSM_NO_PATH)
				return port_num;
		}
	}

	/*
	   Either no existing mcast paths reach this port or we are
	   ignoring existing paths.

	   Determine the best multicast path to the target.  Note that this
	   algorithm is slightly different from the one used for unicast route
	   recommendation.  In this case (multicast), we must NOT
	   perform any sort of load balancing.  We MUST take the FIRST
	   port found that has <= the lowest hop count path.  This prevents
	   more than one multicast path to the same remote switch which
	   prevents a multicast loop.  Multicast loops are bad since the same
	   multicast packet will go around and around, inevitably creating
	   a black hole that will destroy the Earth in a firey conflagration.
	 */
	least_hops = osm_switch_get_least_hops(p_sw, base_lid);
	if (least_hops == OSM_NO_PATH)
		return OSM_NO_PATH;
	for (port_num = 1; port_num < num_ports; port_num++)
		if (osm_switch_get_hop_count(p_sw, base_lid, port_num) ==
		    least_hops)
			break;

	CL_ASSERT(port_num < num_ports);
	return port_num;
}
uint8_t osm_switch_recommend_path(IN const osm_switch_t * p_sw,
				  IN osm_port_t * p_port, IN uint16_t lid_ho,
				  IN unsigned start_from,
				  IN boolean_t ignore_existing,
				  IN boolean_t routing_for_lmc,
				  IN boolean_t dor,
				  IN boolean_t port_shifting,
				  IN uint32_t scatter_ports)
{
	/*
	   We support an enhanced LMC aware routing mode:
	   In the case of LMC > 0, we can track the remote side
	   system and node for all of the lids of the target
	   and try and avoid routing again through the same
	   system / node.

	   Assume if routing_for_lmc is true that this procedure was
	   provided the tracking array and counter via p_port->priv,
	   and we can conduct this algorithm.
	 */
	uint16_t base_lid;
	uint8_t hops;
	uint8_t least_hops;
	uint8_t port_num;
	uint8_t num_ports;
	uint32_t least_paths = 0xFFFFFFFF;
	unsigned i;
	/*
	   The follwing will track the least paths if the
	   route should go through a new system/node
	 */
	uint32_t least_paths_other_sys = 0xFFFFFFFF;
	uint32_t least_paths_other_nodes = 0xFFFFFFFF;
	uint32_t least_forwarded_to = 0xFFFFFFFF;
	uint32_t check_count;
	uint8_t best_port = 0;
	/*
	   These vars track the best port if it connects to
	   not used system/node.
	 */
	uint8_t best_port_other_sys = 0;
	uint8_t best_port_other_node = 0;
	boolean_t port_found = FALSE;
	osm_physp_t *p_physp;
	osm_physp_t *p_rem_physp;
	osm_node_t *p_rem_node;
	osm_node_t *p_rem_node_first = NULL;
	struct osm_remote_node *p_remote_guid = NULL;
	struct osm_remote_node null_remote_node = {NULL, 0, 0};
	struct switch_port_path port_paths[IB_NODE_NUM_PORTS_MAX];
	unsigned int port_paths_total_paths = 0;
	unsigned int port_paths_count = 0;
	uint8_t scatter_possible_ports[IB_NODE_NUM_PORTS_MAX];
	unsigned int scatter_possible_ports_count = 0;
	int found_sys_guid = 0;
	int found_node_guid = 0;

	CL_ASSERT(lid_ho > 0);

	if (p_port->p_node->sw) {
		if (p_port->p_node->sw == p_sw)
			return 0;
		base_lid = osm_port_get_base_lid(p_port);
	} else {
		p_physp = p_port->p_physp;
		if (!p_physp || !p_physp->p_remote_physp ||
		    !p_physp->p_remote_physp->p_node->sw)
			return OSM_NO_PATH;

		if (p_physp->p_remote_physp->p_node->sw == p_sw)
			return p_physp->p_remote_physp->port_num;
		base_lid =
		    osm_node_get_base_lid(p_physp->p_remote_physp->p_node, 0);
	}
	base_lid = cl_ntoh16(base_lid);

	num_ports = p_sw->num_ports;

	least_hops = osm_switch_get_least_hops(p_sw, base_lid);
	if (least_hops == OSM_NO_PATH)
		return OSM_NO_PATH;

	/*
	   First, inquire with the forwarding table for an existing
	   route.  If one is found, honor it unless:
	   1. the ignore existing flag is set.
	   2. the physical port is not a valid one or not healthy
	   3. the physical port has a remote port (the link is up)
	   4. the port has min-hops to the target (avoid loops)
	 */
	if (!ignore_existing) {
		port_num = osm_switch_get_port_by_lid(p_sw, lid_ho);

		if (port_num != OSM_NO_PATH) {
			CL_ASSERT(port_num < num_ports);

			p_physp =
			    osm_node_get_physp_ptr(p_sw->p_node, port_num);
			/*
			   Don't be too trusting of the current forwarding table!
			   Verify that the port number is legal and that the
			   LID is reachable through this port.
			 */
			if (p_physp && osm_physp_is_healthy(p_physp) &&
			    osm_physp_get_remote(p_physp)) {
				hops =
				    osm_switch_get_hop_count(p_sw, base_lid,
							     port_num);
				/*
				   If we aren't using pre-defined user routes
				   function, then we need to make sure that the
				   current path is the minimum one. In case of
				   having such a user function - this check will
				   not be done, and the old routing will be used.
				   Note: This means that it is the user's job to
				   clean all data in the forwarding tables that
				   he wants to be overridden by the minimum
				   hop function.
				 */
				if (hops == least_hops)
					return port_num;
			}
		}
	}

	/*
	   This algorithm selects a port based on a static load balanced
	   selection across equal hop-count ports.
	   There is lots of room for improved sophistication here,
	   possibly guided by user configuration info.
	 */

	/*
	   OpenSM routing is "local" - not considering a full lid to lid
	   path. As such we can not guarantee a path will not loop if we
	   do not always follow least hops.
	   So we must abort if not least hops.
	 */

	/* port number starts with one and num_ports is 1 + num phys ports */
	for (i = start_from; i < start_from + num_ports; i++) {
		port_num = osm_switch_get_dimn_port(p_sw, i % num_ports);
		if (!port_num ||
		    osm_switch_get_hop_count(p_sw, base_lid, port_num) !=
		    least_hops)
			continue;

		/* let us make sure it is not down or unhealthy */
		p_physp = osm_node_get_physp_ptr(p_sw->p_node, port_num);
		if (!p_physp || !osm_physp_is_healthy(p_physp) ||
		    /*
		       we require all - non sma ports to be linked
		       to be routed through
		     */
		    !osm_physp_get_remote(p_physp))
			continue;

		/*
		   We located a least-hop port, possibly one of many.
		   For this port, check the running total count of
		   the number of paths through this port.  Select
		   the port routing the least number of paths.
		 */
		check_count =
		    osm_port_prof_path_count_get(&p_sw->p_prof[port_num]);


		if (dor) {
			/* Get the Remote Node */
			p_rem_physp = osm_physp_get_remote(p_physp);
			p_rem_node = osm_physp_get_node_ptr(p_rem_physp);
			/* use the first dimension, but spread traffic
			 * out among the group of ports representing
			 * that dimension */
			if (!p_rem_node_first)
				p_rem_node_first = p_rem_node;
			else if (p_rem_node != p_rem_node_first)
				continue;
			if (routing_for_lmc) {
				struct osm_remote_guids_count *r = p_port->priv;
				uint8_t rem_port = osm_physp_get_port_num(p_rem_physp);
				unsigned int j;

				for (j = 0; j < r->count; j++) {
					p_remote_guid = &r->guids[j];
					if ((p_remote_guid->node == p_rem_node)
					    && (p_remote_guid->port == rem_port))
						break;
				}
				if (j == r->count)
					p_remote_guid = &null_remote_node;
			}
		/*
		   Advanced LMC routing requires tracking of the
		   best port by the node connected to the other side of
		   it.
		 */
		} else if (routing_for_lmc) {
			/* Is the sys guid already used ? */
			p_remote_guid = switch_find_sys_guid_count(p_sw,
								   p_port->priv,
								   port_num);

			/* If not update the least hops for this case */
			if (!p_remote_guid) {
				if (check_count < least_paths_other_sys) {
					least_paths_other_sys = check_count;
					best_port_other_sys = port_num;
					least_forwarded_to = 0;
				}
				found_sys_guid = 0;
			} else {	/* same sys found - try node */


				/* Else is the node guid already used ? */
				p_remote_guid = switch_find_node_guid_count(p_sw,
									    p_port->priv,
									    port_num);

				/* If not update the least hops for this case */
				if (!p_remote_guid
				    && check_count < least_paths_other_nodes) {
					least_paths_other_nodes = check_count;
					best_port_other_node = port_num;
					least_forwarded_to = 0;
				}
				/* else prior sys and node guid already used */

				if (!p_remote_guid)
					found_node_guid = 0;
				else
					found_node_guid = 1;
				found_sys_guid = 1;
			}	/* same sys found */
		}

		port_paths[port_paths_count].port_num = port_num;
		port_paths[port_paths_count].path_count = check_count;
		if (routing_for_lmc) {
			port_paths[port_paths_count].found_sys_guid = found_sys_guid;
			port_paths[port_paths_count].found_node_guid = found_node_guid;
		}
		if (routing_for_lmc && p_remote_guid)
			port_paths[port_paths_count].forwarded_to = p_remote_guid->forwarded_to;
		else
			port_paths[port_paths_count].forwarded_to = 0;
		port_paths_total_paths += check_count;
		port_paths_count++;

		/* routing for LMC mode */
		/*
		   the count is min but also lower then the max subscribed
		 */
		if (check_count < least_paths) {
			port_found = TRUE;
			best_port = port_num;
			least_paths = check_count;
			scatter_possible_ports_count = 0;
			scatter_possible_ports[scatter_possible_ports_count++] = port_num;
			if (routing_for_lmc
			    && p_remote_guid
			    && p_remote_guid->forwarded_to < least_forwarded_to)
				least_forwarded_to = p_remote_guid->forwarded_to;
		} else if (scatter_ports
			   && check_count == least_paths) {
			scatter_possible_ports[scatter_possible_ports_count++] = port_num;
		} else if (routing_for_lmc
			   && p_remote_guid
			   && check_count == least_paths
			   && p_remote_guid->forwarded_to < least_forwarded_to) {
			least_forwarded_to = p_remote_guid->forwarded_to;
			best_port = port_num;
		}
	}

	if (port_found == FALSE)
		return OSM_NO_PATH;

	if (port_shifting && port_paths_count) {
		/* In the port_paths[] array, we now have all the ports that we
		 * can route out of.  Using some shifting math below, possibly
		 * select a different one so that lids won't align in LFTs
		 *
		 * If lmc > 0, we need to loop through these ports to find the
		 * least_forwarded_to port, best_port_other_sys, and
		 * best_port_other_node just like before but through the different
		 * ordering.
		 */

		least_paths = 0xFFFFFFFF;
		least_paths_other_sys = 0xFFFFFFFF;
		least_paths_other_nodes = 0xFFFFFFFF;
	        least_forwarded_to = 0xFFFFFFFF;
		best_port = 0;
		best_port_other_sys = 0;
		best_port_other_node = 0;

		for (i = 0; i < port_paths_count; i++) {
			unsigned int idx;

			idx = (port_paths_total_paths/port_paths_count + i) % port_paths_count;

			if (routing_for_lmc) {
				if (!port_paths[idx].found_sys_guid
				    && port_paths[idx].path_count < least_paths_other_sys) {
					least_paths_other_sys = port_paths[idx].path_count;
					best_port_other_sys = port_paths[idx].port_num;
					least_forwarded_to = 0;
				}
				else if (!port_paths[idx].found_node_guid
					 && port_paths[idx].path_count < least_paths_other_nodes) {
					least_paths_other_nodes = port_paths[idx].path_count;
					best_port_other_node = port_paths[idx].port_num;
					least_forwarded_to = 0;
				}
			}

			if (port_paths[idx].path_count < least_paths) {
				best_port = port_paths[idx].port_num;
				least_paths = port_paths[idx].path_count;
				if (routing_for_lmc
				    && (port_paths[idx].found_sys_guid
					|| port_paths[idx].found_node_guid)
				    && port_paths[idx].forwarded_to < least_forwarded_to)
					least_forwarded_to = port_paths[idx].forwarded_to;
			}
			else if (routing_for_lmc
				 && (port_paths[idx].found_sys_guid
				     || port_paths[idx].found_node_guid)
				 && port_paths[idx].path_count == least_paths
				 && port_paths[idx].forwarded_to < least_forwarded_to) {
				least_forwarded_to = port_paths[idx].forwarded_to;
				best_port = port_paths[idx].port_num;
			}

		}
	}

	/*
	   if we are in enhanced routing mode and the best port is not
	   the local port 0
	 */
	if (routing_for_lmc && best_port && !scatter_ports) {
		/* Select the least hop port of the non used sys first */
		if (best_port_other_sys)
			best_port = best_port_other_sys;
		else if (best_port_other_node)
			best_port = best_port_other_node;
	} else if (scatter_ports) {
		/*
		 * There is some danger that this random could "rebalance" the routes
		 * every time, to combat this there is a global srandom that
		 * occurs at the start of every sweep.
		 */
		unsigned int idx = random() % scatter_possible_ports_count;
		best_port = scatter_possible_ports[idx];
	}
	return best_port;
}
Beispiel #6
0
uint8_t
osm_switch_recommend_path(IN const osm_switch_t * const p_sw,
			  IN osm_port_t * p_port,
			  IN const uint16_t lid_ho,
			  IN unsigned start_from,
			  IN const boolean_t ignore_existing,
			  IN const boolean_t dor)
{
	/*
	   We support an enhanced LMC aware routing mode:
	   In the case of LMC > 0, we can track the remote side
	   system and node for all of the lids of the target
	   and try and avoid routing again through the same
	   system / node.

	   If this procedure is provided with the tracking array
	   and counter we can conduct this algorithm.
	 */
	boolean_t routing_for_lmc = (p_port->priv != NULL);
	uint16_t base_lid;
	uint8_t hops;
	uint8_t least_hops;
	uint8_t port_num;
	uint8_t num_ports;
	uint32_t least_paths = 0xFFFFFFFF;
	unsigned i;
	/*
	   The follwing will track the least paths if the
	   route should go through a new system/node
	 */
	uint32_t least_paths_other_sys = 0xFFFFFFFF;
	uint32_t least_paths_other_nodes = 0xFFFFFFFF;
	uint32_t least_forwarded_to = 0xFFFFFFFF;
	uint32_t check_count;
	uint8_t best_port = 0;
	/*
	   These vars track the best port if it connects to
	   not used system/node.
	 */
	uint8_t best_port_other_sys = 0;
	uint8_t best_port_other_node = 0;
	boolean_t port_found = FALSE;
	osm_physp_t *p_physp;
	osm_physp_t *p_rem_physp;
	osm_node_t *p_rem_node;
	osm_node_t *p_rem_node_first = NULL;
	struct osm_remote_node *p_remote_guid = NULL;

	CL_ASSERT(lid_ho > 0);

	if (p_port->p_node->sw) {
		if (p_port->p_node->sw == p_sw)
			return 0;
		base_lid = osm_port_get_base_lid(p_port);
	} else {
		p_physp = p_port->p_physp;
		if (!p_physp || !p_physp->p_remote_physp ||
		    !p_physp->p_remote_physp->p_node->sw)
			return OSM_NO_PATH;

		if (p_physp->p_remote_physp->p_node->sw == p_sw)
			return p_physp->p_remote_physp->port_num;
		base_lid =
		    osm_node_get_base_lid(p_physp->p_remote_physp->p_node, 0);
	}
	base_lid = cl_ntoh16(base_lid);

	num_ports = p_sw->num_ports;

	least_hops = osm_switch_get_least_hops(p_sw, base_lid);
	if (least_hops == OSM_NO_PATH)
		return (OSM_NO_PATH);

	/*
	   First, inquire with the forwarding table for an existing
	   route.  If one is found, honor it unless:
	   1. the ignore existing flag is set.
	   2. the physical port is not a valid one or not healthy
	   3. the physical port has a remote port (the link is up)
	   4. the port has min-hops to the target (avoid loops)
	 */
	if (!ignore_existing) {
		port_num = osm_switch_get_port_by_lid(p_sw, lid_ho);

		if (port_num != OSM_NO_PATH) {
			CL_ASSERT(port_num < num_ports);

			p_physp =
			    osm_node_get_physp_ptr(p_sw->p_node, port_num);
			/*
			   Don't be too trusting of the current forwarding table!
			   Verify that the port number is legal and that the
			   LID is reachable through this port.
			 */
			if (p_physp && osm_physp_is_healthy(p_physp) &&
			    osm_physp_get_remote(p_physp)) {
				hops =
				    osm_switch_get_hop_count(p_sw, base_lid,
							     port_num);
				/*
				   If we aren't using pre-defined user routes
				   function, then we need to make sure that the
				   current path is the minimum one. In case of
				   having such a user function - this check will
				   not be done, and the old routing will be used.
				   Note: This means that it is the user's job to
				   clean all data in the forwarding tables that
				   he wants to be overridden by the minimum
				   hop function.
				 */
				if (hops == least_hops)
					return (port_num);
			}
		}
	}

	/*
	   This algorithm selects a port based on a static load balanced
	   selection across equal hop-count ports.
	   There is lots of room for improved sophistication here,
	   possibly guided by user configuration info.
	 */

	/*
	   OpenSM routing is "local" - not considering a full lid to lid
	   path. As such we can not guarantee a path will not loop if we
	   do not always follow least hops.
	   So we must abort if not least hops.
	 */

	/* port number starts with one and num_ports is 1 + num phys ports */
	for (i = start_from; i < start_from + num_ports; i++) {
		port_num = i%num_ports;
		if (!port_num ||
		    osm_switch_get_hop_count(p_sw, base_lid, port_num) !=
		    least_hops)
			continue;

		/* let us make sure it is not down or unhealthy */
		p_physp = osm_node_get_physp_ptr(p_sw->p_node, port_num);
		if (!p_physp || !osm_physp_is_healthy(p_physp) ||
		    /*
		       we require all - non sma ports to be linked
		       to be routed through
		     */
		    !osm_physp_get_remote(p_physp))
			continue;

		/*
		   We located a least-hop port, possibly one of many.
		   For this port, check the running total count of
		   the number of paths through this port.  Select
		   the port routing the least number of paths.
		 */
		check_count =
		    osm_port_prof_path_count_get(&p_sw->p_prof[port_num]);

		/*
		   Advanced LMC routing requires tracking of the
		   best port by the node connected to the other side of
		   it.
		 */
		if (routing_for_lmc) {
			/* Is the sys guid already used ? */
			p_remote_guid = osm_switch_find_sys_guid_count(p_sw,
								       p_port->priv,
								       port_num);

			/* If not update the least hops for this case */
			if (!p_remote_guid) {
				if (check_count < least_paths_other_sys) {
					least_paths_other_sys = check_count;
					best_port_other_sys = port_num;
					least_forwarded_to = 0;
				}
			} else {	/* same sys found - try node */
				/* Else is the node guid already used ? */
				p_remote_guid = osm_switch_find_node_guid_count(p_sw,
										p_port->priv,
										port_num);

				/* If not update the least hops for this case */
				if (!p_remote_guid
				    && check_count < least_paths_other_nodes) {
					least_paths_other_nodes = check_count;
					best_port_other_node = port_num;
					least_forwarded_to = 0;
				}
				/* else prior sys and node guid already used */

			}	/* same sys found */
		}

		/* routing for LMC mode */
		/*
		   the count is min but also lower then the max subscribed
		 */
		if (check_count < least_paths) {
			if (dor) {
				/* Get the Remote Node */
				p_rem_physp = osm_physp_get_remote(p_physp);
				p_rem_node =
				    osm_physp_get_node_ptr(p_rem_physp);
				/* use the first dimension, but spread
				 * traffic out among the group of ports
				 * representing that dimension */
				if (port_found) {
					if (p_rem_node != p_rem_node_first)
						continue;
				} else
					p_rem_node_first = p_rem_node;
			}
			port_found = TRUE;
			best_port = port_num;
			least_paths = check_count;
			if (routing_for_lmc
			    && p_remote_guid
			    && p_remote_guid->forwarded_to < least_forwarded_to)
				least_forwarded_to = p_remote_guid->forwarded_to;
		} else if (routing_for_lmc
			   && p_remote_guid
			   && check_count == least_paths
			   && p_remote_guid->forwarded_to < least_forwarded_to) {
			least_forwarded_to = p_remote_guid->forwarded_to;
			best_port = port_num;
		}
	}

	if (port_found == FALSE)
		return (OSM_NO_PATH);

	/*
	   if we are in enhanced routing mode and the best port is not
	   the local port 0
	 */
	if (routing_for_lmc && best_port) {
		/* Select the least hop port of the non used sys first */
		if (best_port_other_sys)
			best_port = best_port_other_sys;
		else if (best_port_other_node)
			best_port = best_port_other_node;
	}

	return (best_port);
}
Beispiel #7
0
static void dump_ucast_routes(cl_map_item_t * item, FILE * file, void *cxt)
{
	const osm_node_t *p_node;
	osm_port_t *p_port;
	uint8_t port_num;
	uint8_t num_hops;
	uint8_t best_hops;
	uint8_t best_port;
	uint16_t max_lid_ho;
	uint16_t lid_ho, base_lid;
	boolean_t direct_route_exists = FALSE;
	boolean_t dor;
	osm_switch_t *p_sw = (osm_switch_t *) item;
	osm_opensm_t *p_osm = cxt;

	p_node = p_sw->p_node;

	max_lid_ho = p_sw->max_lid_ho;

	fprintf(file, "dump_ucast_routes: "
		"Switch 0x%016" PRIx64 "\nLID    : Port : Hops : Optimal\n",
		cl_ntoh64(osm_node_get_node_guid(p_node)));

	dor = (p_osm->routing_engine_used &&
	       p_osm->routing_engine_used->type == OSM_ROUTING_ENGINE_TYPE_DOR);

	for (lid_ho = 1; lid_ho <= max_lid_ho; lid_ho++) {
		fprintf(file, "0x%04X : ", lid_ho);

		p_port = osm_get_port_by_lid_ho(&p_osm->subn, lid_ho);
		if (!p_port) {
			fprintf(file, "UNREACHABLE\n");
			continue;
		}

		port_num = osm_switch_get_port_by_lid(p_sw, lid_ho,
						      OSM_NEW_LFT);
		if (port_num == OSM_NO_PATH) {
			/*
			   This may occur if there are 'holes' in the existing
			   LID assignments.  Running SM with --reassign_lids
			   will reassign and compress the LID range.  The
			   subnet should work fine either way.
			 */
			fprintf(file, "UNREACHABLE\n");
			continue;
		}
		/*
		   Switches can lie about which port routes a given
		   lid due to a recent reconfiguration of the subnet.
		   Therefore, ensure that the hop count is better than
		   OSM_NO_PATH.
		 */
		if (p_port->p_node->sw) {
			/* Target LID is switch.
			   Get its base lid and check hop count for this base LID only. */
			base_lid = osm_node_get_base_lid(p_port->p_node, 0);
			base_lid = cl_ntoh16(base_lid);
			num_hops =
			    osm_switch_get_hop_count(p_sw, base_lid, port_num);
		} else {
			/* Target LID is not switch (CA or router).
			   Check if we have route to this target from current switch. */
			num_hops =
			    osm_switch_get_hop_count(p_sw, lid_ho, port_num);
			if (num_hops != OSM_NO_PATH) {
				direct_route_exists = TRUE;
				base_lid = lid_ho;
			} else {
				osm_physp_t *p_physp = p_port->p_physp;

				if (!p_physp || !p_physp->p_remote_physp ||
				    !p_physp->p_remote_physp->p_node->sw)
					num_hops = OSM_NO_PATH;
				else {
					base_lid =
					    osm_node_get_base_lid(p_physp->
								  p_remote_physp->
								  p_node, 0);
					base_lid = cl_ntoh16(base_lid);
					num_hops =
					    p_physp->p_remote_physp->p_node->
					    sw ==
					    p_sw ? 0 :
					    osm_switch_get_hop_count(p_sw,
								     base_lid,
								     port_num);
				}
			}
		}

		if (num_hops == OSM_NO_PATH) {
			fprintf(file, "%03u  : HOPS UNKNOWN\n", port_num);
			continue;
		}

		best_hops = osm_switch_get_least_hops(p_sw, base_lid);
		if (!p_port->p_node->sw && !direct_route_exists) {
			best_hops++;
			num_hops++;
		}

		fprintf(file, "%03u  : %02u   : ", port_num, num_hops);

		if (best_hops == num_hops)
			fprintf(file, "yes");
		else {
			/* No LMC Optimization */
			best_port = osm_switch_recommend_path(p_sw, p_port,
							      lid_ho, 1, TRUE,
							      FALSE, dor,
							      p_osm->subn.opt.port_shifting,
							      p_osm->subn.opt.scatter_ports,
							      OSM_NEW_LFT);
			fprintf(file, "No %u hop path possible via port %u!",
				best_hops, best_port);
		}

		fprintf(file, "\n");
	}
}
Beispiel #8
0
/**********************************************************************
 * This function does the bfs of min hop table calculation by guid index
 * as a starting point.
 **********************************************************************/
static int updn_bfs_by_node(IN osm_log_t * p_log, IN osm_subn_t * p_subn,
			    IN osm_switch_t * p_sw)
{
	uint8_t pn, pn_rem;
	cl_qlist_t list;
	uint16_t lid;
	struct updn_node *u;
	updn_switch_dir_t next_dir, current_dir;

	OSM_LOG_ENTER(p_log);

	lid = osm_node_get_base_lid(p_sw->p_node, 0);
	lid = cl_ntoh16(lid);
	osm_switch_set_hops(p_sw, lid, 0, 0);

	OSM_LOG(p_log, OSM_LOG_DEBUG,
		"Starting from switch - port GUID 0x%" PRIx64 " lid %u\n",
		cl_ntoh64(p_sw->p_node->node_info.port_guid), lid);

	u = p_sw->priv;
	u->dir = UP;

	/* Update list with the new element */
	cl_qlist_init(&list);
	cl_qlist_insert_tail(&list, &u->list);

	/* BFS the list till no next element */
	while (!cl_is_qlist_empty(&list)) {
		u = (struct updn_node *)cl_qlist_remove_head(&list);
		u->visited = 0;	/* cleanup */
		current_dir = u->dir;
		/* Go over all ports of the switch and find unvisited remote nodes */
		for (pn = 1; pn < u->sw->num_ports; pn++) {
			osm_node_t *p_remote_node;
			struct updn_node *rem_u;
			uint8_t current_min_hop, remote_min_hop,
			    set_hop_return_value;
			osm_switch_t *p_remote_sw;

			p_remote_node =
			    osm_node_get_remote_node(u->sw->p_node, pn,
						     &pn_rem);
			/* If no remote node OR remote node is not a SWITCH
			   continue to next pn */
			if (!p_remote_node || !p_remote_node->sw)
				continue;
			/* Fetch remote guid only after validation of remote node */
			p_remote_sw = p_remote_node->sw;
			rem_u = p_remote_sw->priv;
			/* Decide which direction to mark it (UP/DOWN) */
			next_dir = updn_get_dir(u->rank, rem_u->rank,
						u->id, rem_u->id);

			/* Check if this is a legal step : the only illegal step is going
			   from DOWN to UP */
			if ((current_dir == DOWN) && (next_dir == UP)) {
				OSM_LOG(p_log, OSM_LOG_DEBUG,
					"Avoiding move from 0x%016" PRIx64
					" to 0x%016" PRIx64 "\n",
					cl_ntoh64(osm_node_get_node_guid(u->sw->p_node)),
					cl_ntoh64(osm_node_get_node_guid(p_remote_node)));
				/* Illegal step */
				continue;
			}
			/* Set MinHop value for the current lid */
			current_min_hop = osm_switch_get_least_hops(u->sw, lid);
			/* Check hop count if better insert into list && update
			   the remote node Min Hop Table */
			remote_min_hop =
			    osm_switch_get_hop_count(p_remote_sw, lid, pn_rem);
			if (current_min_hop + 1 < remote_min_hop) {
				set_hop_return_value =
				    osm_switch_set_hops(p_remote_sw, lid,
							pn_rem,
							current_min_hop + 1);
				if (set_hop_return_value) {
					OSM_LOG(p_log, OSM_LOG_ERROR, "ERR AA01: "
						"Invalid value returned from set min hop is: %d\n",
						set_hop_return_value);
				}
				/* Check if remote port has already been visited */
				if (!rem_u->visited) {
					/* Insert updn_switch item into the list */
					rem_u->dir = next_dir;
					rem_u->visited = 1;
					cl_qlist_insert_tail(&list,
							     &rem_u->list);
				}
			}
		}
	}

	OSM_LOG_EXIT(p_log);
	return 0;
}