1============== 2Page migration 3============== 4 5Page migration allows moving the physical location of pages between 6nodes in a NUMA system while the process is running. This means that the 7virtual addresses that the process sees do not change. However, the 8system rearranges the physical location of those pages. 9 10Also see Documentation/mm/hmm.rst for migrating pages to or from device 11private memory. 12 13The main intent of page migration is to reduce the latency of memory accesses 14by moving pages near to the processor where the process accessing that memory 15is running. 16 17Page migration allows a process to manually relocate the node on which its 18pages are located through the MF_MOVE and MF_MOVE_ALL options while setting 19a new memory policy via mbind(). The pages of a process can also be relocated 20from another process using the sys_migrate_pages() function call. The 21migrate_pages() function call takes two sets of nodes and moves pages of a 22process that are located on the from nodes to the destination nodes. 23Page migration functions are provided by the numactl package by Andi Kleen 24(a version later than 0.9.3 is required. Get it from 25https://github.com/numactl/numactl.git). numactl provides libnuma 26which provides an interface similar to other NUMA functionality for page 27migration. cat ``/proc/<pid>/numa_maps`` allows an easy review of where the 28pages of a process are located. See also the numa_maps documentation in the 29proc(5) man page. 30 31Manual migration is useful if for example the scheduler has relocated 32a process to a processor on a distant node. A batch scheduler or an 33administrator may detect the situation and move the pages of the process 34nearer to the new processor. The kernel itself only provides 35manual page migration support. Automatic page migration may be implemented 36through user space processes that move pages. A special function call 37"move_pages" allows the moving of individual pages within a process. 38For example, A NUMA profiler may obtain a log showing frequent off-node 39accesses and may use the result to move pages to more advantageous 40locations. 41 42Larger installations usually partition the system using cpusets into 43sections of nodes. Paul Jackson has equipped cpusets with the ability to 44move pages when a task is moved to another cpuset (See 45:ref:`CPUSETS <cpusets>`). 46Cpusets allow the automation of process locality. If a task is moved to 47a new cpuset then also all its pages are moved with it so that the 48performance of the process does not sink dramatically. Also the pages 49of processes in a cpuset are moved if the allowed memory nodes of a 50cpuset are changed. 51 52Page migration allows the preservation of the relative location of pages 53within a group of nodes for all migration techniques which will preserve a 54particular memory allocation pattern generated even after migrating a 55process. This is necessary in order to preserve the memory latencies. 56Processes will run with similar performance after migration. 57 58Page migration occurs in several steps. First a high level 59description for those trying to use migrate_pages() from the kernel 60(for userspace usage see the Andi Kleen's numactl package mentioned above) 61and then a low level description of how the low level details work. 62 63In kernel use of migrate_pages() 64================================ 65 661. Remove pages from the LRU. 67 68 Lists of pages to be migrated are generated by scanning over 69 pages and moving them into lists. This is done by 70 calling isolate_lru_page(). 71 Calling isolate_lru_page() increases the references to the page 72 so that it cannot vanish while the page migration occurs. 73 It also prevents the swapper or other scans from encountering 74 the page. 75 762. We need to have a function of type new_page_t that can be 77 passed to migrate_pages(). This function should figure out 78 how to allocate the correct new page given the old page. 79 803. The migrate_pages() function is called which attempts 81 to do the migration. It will call the function to allocate 82 the new page for each page that is considered for 83 moving. 84 85How migrate_pages() works 86========================= 87 88migrate_pages() does several passes over its list of pages. A page is moved 89if all references to a page are removable at the time. The page has 90already been removed from the LRU via isolate_lru_page() and the refcount 91is increased so that the page cannot be freed while page migration occurs. 92 93Steps: 94 951. Lock the page to be migrated. 96 972. Ensure that writeback is complete. 98 993. Lock the new page that we want to move to. It is locked so that accesses to 100 this (not yet up-to-date) page immediately block while the move is in progress. 101 1024. All the page table references to the page are converted to migration 103 entries. This decreases the mapcount of a page. If the resulting 104 mapcount is not zero then we do not migrate the page. All user space 105 processes that attempt to access the page will now wait on the page lock 106 or wait for the migration page table entry to be removed. 107 1085. The i_pages lock is taken. This will cause all processes trying 109 to access the page via the mapping to block on the spinlock. 110 1116. The refcount of the page is examined and we back out if references remain. 112 Otherwise, we know that we are the only one referencing this page. 113 1147. The radix tree is checked and if it does not contain the pointer to this 115 page then we back out because someone else modified the radix tree. 116 1178. The new page is prepped with some settings from the old page so that 118 accesses to the new page will discover a page with the correct settings. 119 1209. The radix tree is changed to point to the new page. 121 12210. The reference count of the old page is dropped because the address space 123 reference is gone. A reference to the new page is established because 124 the new page is referenced by the address space. 125 12611. The i_pages lock is dropped. With that lookups in the mapping 127 become possible again. Processes will move from spinning on the lock 128 to sleeping on the locked new page. 129 13012. The page contents are copied to the new page. 131 13213. The remaining page flags are copied to the new page. 133 13414. The old page flags are cleared to indicate that the page does 135 not provide any information anymore. 136 13715. Queued up writeback on the new page is triggered. 138 13916. If migration entries were inserted into the page table, then replace them 140 with real ptes. Doing so will enable access for user space processes not 141 already waiting for the page lock. 142 14317. The page locks are dropped from the old and new page. 144 Processes waiting on the page lock will redo their page faults 145 and will reach the new page. 146 14718. The new page is moved to the LRU and can be scanned by the swapper, 148 etc. again. 149 150Non-LRU page migration 151====================== 152 153Although migration originally aimed for reducing the latency of memory 154accesses for NUMA, compaction also uses migration to create high-order 155pages. For compaction purposes, it is also useful to be able to move 156non-LRU pages, such as zsmalloc and virtio-balloon pages. 157 158If a driver wants to make its pages movable, it should define a struct 159movable_operations. It then needs to call __SetPageMovable() on each 160page that it may be able to move. This uses the ``page->mapping`` field, 161so this field is not available for the driver to use for other purposes. 162 163Monitoring Migration 164===================== 165 166The following events (counters) can be used to monitor page migration. 167 1681. PGMIGRATE_SUCCESS: Normal page migration success. Each count means that a 169 page was migrated. If the page was a non-THP and non-hugetlb page, then 170 this counter is increased by one. If the page was a THP or hugetlb, then 171 this counter is increased by the number of THP or hugetlb subpages. 172 For example, migration of a single 2MB THP that has 4KB-size base pages 173 (subpages) will cause this counter to increase by 512. 174 1752. PGMIGRATE_FAIL: Normal page migration failure. Same counting rules as for 176 PGMIGRATE_SUCCESS, above: this will be increased by the number of subpages, 177 if it was a THP or hugetlb. 178 1793. THP_MIGRATION_SUCCESS: A THP was migrated without being split. 180 1814. THP_MIGRATION_FAIL: A THP could not be migrated nor it could be split. 182 1835. THP_MIGRATION_SPLIT: A THP was migrated, but not as such: first, the THP had 184 to be split. After splitting, a migration retry was used for it's sub-pages. 185 186THP_MIGRATION_* events also update the appropriate PGMIGRATE_SUCCESS or 187PGMIGRATE_FAIL events. For example, a THP migration failure will cause both 188THP_MIGRATION_FAIL and PGMIGRATE_FAIL to increase. 189 190Christoph Lameter, May 8, 2006. 191Minchan Kim, Mar 28, 2016. 192 193.. kernel-doc:: include/linux/migrate.h 194