Lines Matching defs:nodes
1674 * between these nodes are slowed down, to allow things to settle down.
1683 /* Handle placement on systems where not all nodes are directly connected. */
1691 * All nodes are directly connected, and the same distance
1701 * which should be ok given the number of nodes rarely exceeds 8.
1708 * The furthest away nodes in the system are not interesting
1716 * of nodes, and move tasks towards the group with the most
1717 * memory accesses. When comparing two nodes at distance
1718 * "hoplimit", only nodes closer by than "hoplimit" are part
1719 * of each group. Skip other nodes.
1724 /* Add up the faults from nearby nodes. */
1732 * no fixed "groups of nodes". Instead, nodes that are not
1734 * nodes; a numa_group can occupy any set of nodes.
1753 * evenly spread out between numa nodes.
2546 * Look at other nodes in these cases:
2549 * multiple NUMA nodes; in order to better consolidate the group,
2565 /* Only consider nodes where both task and groups benefit */
2579 * If the task is part of a workload that spans multiple NUMA nodes,
2580 * and is migrating into one of the workload's active nodes, remember
2642 * Find out how many nodes the workload is actively running on. Do this by
2643 * tracking the nodes from which NUMA hinting faults are triggered. This can
2644 * be different from the set of nodes where the workload's memory is currently
2794 nodemask_t nodes;
2797 /* Direct connections between all NUMA nodes. */
2803 * scores nodes according to the number of NUMA hinting faults on
2804 * both the node itself, and on nearby nodes.
2828 * inside the highest scoring group of nodes. The nodemask tricks
2831 nodes = node_states[N_CPU];
2837 /* Are there nodes at this distance from each other? */
2841 for_each_node_mask(a, nodes) {
2847 for_each_node_mask(b, nodes) {
2851 node_clear(b, nodes);
2867 /* Next round, evaluate the nodes within max_group. */
2870 nodes = max_group;
3210 * If a workload spans multiple NUMA nodes, a shared fault that
3211 * occurs wholly within the set of nodes that the workload is