kernel/sched/fair.c

43 #include <linux/memory-tiers.h>
1334 	 * for both memory bandwidth and any shared HT resources.  This is an
1355  * calculated based on the tasks virtual memory size and
1384 	 * Faults_cpu is used to decide whether memory should move
1386 	 * more by CPU use than by memory faults.
1451 	/* Scale the maximum scan period with the amount of shared memory. */
1476 	/* Scale the maximum scan period with the amount of shared memory. */
1529  * The averaged statistics, shared & private, memory & CPU,
1635 		 * memory accesses. When comparing two nodes at distance
1713  * If memory tiering mode is enabled, cpupid of slow memory page is
1725  * For memory tiering mode, if there are enough free pages (more than
1726  * enough watermark defined here) in fast memory node, to take full
1727  * advantage of fast memory capacity, all recently accessed slow
1728  * memory pages will be migrated to fast memory node without
1753  * For memory tiering mode, when page tables are scanned, the scan
1755  * PROT_NONE for slow memory page.  So when the page is accessed, in
1775  * For memory tiering mode, too high promotion/demotion throughput may
1840 	 * The pages in slow memory node should be migrated according
1925 	 * Distribute memory according to CPU & memory use on each node,
1926 	 * with 3/4 hysteresis to avoid unnecessary memory migrations:
2568  * be different from the set of nodes where the workload's memory is currently
2604  * our memory is already on our local node, or if the majority of
2647 		 * Most memory accesses are local. There is no need to
2648 		 * do fast NUMA scanning, since memory is already local.
2656 		 * Most memory accesses are shared with other tasks.
2658 		 * since other tasks may just move the memory elsewhere.
2666 		 * Private memory faults exceed (SLOTS-THRESHOLD)/SLOTS,
2668 		 * NUMA scanning to get the memory moved over.
3088 	 * NUMA faults statistics are unnecessary for the slow memory
3089 	 * node for memory tiering mode.
3490  * Drive the periodic memory faults..
3498 	 * We don't care about NUMA placement if we don't have memory.
3611  * memory. This allows lockless observations without ever seeing the negative
3631  * memory. This allows lockless observations without ever seeing the negative