3  * Memory Migration functionality - linux/mm/migrate.c
7  * Page migration was first developed in the context of the memory hotplug
8  * project. The main authors of the migration code are:
81 	 * compaction threads can race against page migration functions  in isolate_movable_page()
85 	 * being (wrongly) re-isolated while it is under migration,  in isolate_movable_page()
127  * from where they were once taken off for compaction/migration.
237  * Restore a potential migration pte to a working pte entry
259 		/* PMD-mapped THP migration entry */  in remove_migration_pte()
345  * Get rid of all migration entries and replace them by
369  * Something used the pte of a page under migration. We need to
370  * get to the page and wait until migration is finished.
422 		 * If migration entry existed, safe to release vma lock  in migration_entry_wait_huge()
672 	 * migration entries. We can still have PG_anon_exclusive set on an  in folio_migrate_flags()
745  *                    Migration functions
774  * migrate_folio() - Simple folio migration.
899  * buffer_migrate_folio() - Migration function for folios with buffers.
921  * buffer_migrate_folio_norefs() - Migration function for folios with buffers.
974 	 * migration. Writeout may mean we lose the lock and the  in writeout()
976 	 * At this point we know that the migration attempt cannot  in writeout()
991  * Default handling if a filesystem does not provide a migration function.
997 		/* Only writeback folios in full synchronous migration */  in fallback_migrate_folio()
1050 			 * for page migration.  in move_to_new_folio()
1061 		 * isolation step. In that case, we shouldn't try migration.  in move_to_new_folio()
1107  * To record some information during migration, we use unused private
1165 /* Cleanup src folio upon migration success */
1250 		 * Only in the case of a full synchronous migration is it  in migrate_folio_unmap()
1269 	 * of migration. File cache pages are no problem because of page_lock()  in migrate_folio_unmap()
1270 	 * File Caches may use write_page() or lock_page() in migration, then,  in migrate_folio_unmap()
1317 		/* Establish migration ptes */  in migrate_folio_unmap()
1387 	 * If migration is successful, decrease refcount of dst,  in migrate_folio_move()
1424  * Counterpart of unmap_and_move_page() for hugepage migration.
1427  * because there is no race between I/O and migration for hugepage.
1435  * hugepage migration fails without data corruption.
1437  * There is also no race when direct I/O is issued on the page under migration,
1438  * because then pte is replaced with migration swap entry and direct I/O code
1439  * will wait in the page fault for migration to complete.
1542 	 * If migration was not successful and there's a freeing callback,  in unmap_and_move_huge_page()
1629 			 * of hugepage migration like soft offline and memory  in migrate_hugetlbs()
1631 			 * the hugepage is pmd-based or not before kicking migration.  in migrate_hugetlbs()
1670 				 * removed from migration folio list and not  in migrate_hugetlbs()
1769  * We only batch migration if mode == MIGRATE_ASYNC to avoid to wait a
1844 			 * Large folio migration might be unsupported or  in migrate_pages_batch()
1935 				 * removed from migration folio list and not  in migrate_pages_batch()
2030  *		   supplied as the target for the page migration
2034  *			as the target of the folio migration.
2035  * @put_new_folio:	The function used to free target folios if migration
2038  * @mode:		The migration mode that specifies the constraints for
2039  *			folio migration, if any.
2040  * @reason:		The reason for folio migration.
2123 	 * Put the permanent failure folio back to migration list, they  in migrate_pages()
2176 		 * clear __GFP_RECLAIM to make the migration callback  in alloc_migration_target()
2376 			/* The page is successfully queued for migration */  in do_pages_move()
2598  * Returns true if this is a safe migration target node for misplaced NUMA