kernel/sched/topology.c

3  * Scheduler topology setup/handling methods
740  * NUMA topology (first read the regular topology blurb below)
750  * which represents a 4 node ring topology like:
762  * For the above NUMA topology that gives 3 levels:
773  * As can be seen; things don't nicely line up as with the regular topology.
776  * the topology.
803  * Another 'interesting' topology is:
819  * This topology is asymmetric, nodes 1,2 are fully connected, but nodes 0,3
992  * Package topology (also see the load-balance blurb in fair.c)
995  * topology features. By default (default_topology[]) these include:
1010  * denote the ever growing mask of CPUs belonging to that level of topology.
1034  * topology levels, while sched_group moves you sideways through it, at child
1148  * unless there are asymmetries in the topology. If there are asymmetries,
1293  * SD_flags allowed in topology descriptions.
1295  * These flags are purely descriptive of the topology and do not prescribe
1303  * Odd one out, which beside describing the topology has a quirk also
1335 			"wrong sd_flags in topology description\n"))  in sd_init()
1338 	/* Apply detected topology flags */  in sd_init()
1423  * Topology list, bottom-up.
1494  * A system can have three types of NUMA topology:
1499  * The difference between a glueless mesh topology and a backplane
1500  * topology lies in whether communication between not directly
1505  * The type of topology can be discerned with the following tests:
1648 	/* Compute default topology size */  in sched_init_numa()
1657 	 * Copy the default topology bits..  in sched_init_numa()
1716  * sched_numa_find_closest() - given the NUMA topology, find the cpu
1855 			pr_err("BUG: arch topology borken\n");  in build_sched_domain()
1873  * Ensure topology masks are sane, i.e. there are no conflicts (overlaps) for
1874  * any two given CPUs at this (non-NUMA) topology level.
1896 		 * match the topology we're about to build, but that can only  in topology_span_sane()
1934 	 * Examine topology from all CPU's point of views to detect the lowest  in asym_cpu_capacity_level()
2090  * CPU core maps. It is supposed to return 1 if the topology changed