blob: d5fdfd34bbaf4835bd7381b93cdd09774f8b9b01 [file] [log] [blame]
Christoph Lametera48d07a2006-02-01 03:05:38 -08001Page migration
2--------------
3
4Page migration allows the moving of the physical location of pages between
5nodes in a numa system while the process is running. This means that the
6virtual addresses that the process sees do not change. However, the
7system rearranges the physical location of those pages.
8
9The main intend of page migration is to reduce the latency of memory access
10by moving pages near to the processor where the process accessing that memory
11is running.
12
13Page migration allows a process to manually relocate the node on which its
14pages are located through the MF_MOVE and MF_MOVE_ALL options while setting
Christoph Lameterb4fb3762006-03-14 19:50:20 -080015a new memory policy via mbind(). The pages of process can also be relocated
Christoph Lametera48d07a2006-02-01 03:05:38 -080016from another process using the sys_migrate_pages() function call. The
17migrate_pages function call takes two sets of nodes and moves pages of a
18process that are located on the from nodes to the destination nodes.
Christoph Lameterb4fb3762006-03-14 19:50:20 -080019Page migration functions are provided by the numactl package by Andi Kleen
20(a version later than 0.9.3 is required. Get it from
Michael Kerrisk6acb2ec2008-08-15 00:40:18 -070021ftp://oss.sgi.com/www/projects/libnuma/download/). numactl provides libnuma
22which provides an interface similar to other numa functionality for page
23migration. cat /proc/<pid>/numa_maps allows an easy review of where the
24pages of a process are located. See also the numa_maps documentation in the
25proc(5) man page.
Christoph Lametera48d07a2006-02-01 03:05:38 -080026
Christoph Lameterb4fb3762006-03-14 19:50:20 -080027Manual migration is useful if for example the scheduler has relocated
Christoph Lametera48d07a2006-02-01 03:05:38 -080028a process to a processor on a distant node. A batch scheduler or an
29administrator may detect the situation and move the pages of the process
Christoph Lameter742755a2006-06-23 02:03:55 -070030nearer to the new processor. The kernel itself does only provide
31manual page migration support. Automatic page migration may be implemented
32through user space processes that move pages. A special function call
33"move_pages" allows the moving of individual pages within a process.
34A NUMA profiler may f.e. obtain a log showing frequent off node
35accesses and may use the result to move pages to more advantageous
36locations.
Christoph Lametera48d07a2006-02-01 03:05:38 -080037
38Larger installations usually partition the system using cpusets into
39sections of nodes. Paul Jackson has equipped cpusets with the ability to
Christoph Lameterb4fb3762006-03-14 19:50:20 -080040move pages when a task is moved to another cpuset (See ../cpusets.txt).
41Cpusets allows the automation of process locality. If a task is moved to
42a new cpuset then also all its pages are moved with it so that the
43performance of the process does not sink dramatically. Also the pages
44of processes in a cpuset are moved if the allowed memory nodes of a
45cpuset are changed.
Christoph Lametera48d07a2006-02-01 03:05:38 -080046
47Page migration allows the preservation of the relative location of pages
48within a group of nodes for all migration techniques which will preserve a
49particular memory allocation pattern generated even after migrating a
50process. This is necessary in order to preserve the memory latencies.
51Processes will run with similar performance after migration.
52
53Page migration occurs in several steps. First a high level
Christoph Lameterb4fb3762006-03-14 19:50:20 -080054description for those trying to use migrate_pages() from the kernel
55(for userspace usage see the Andi Kleen's numactl package mentioned above)
56and then a low level description of how the low level details work.
Christoph Lametera48d07a2006-02-01 03:05:38 -080057
Christoph Lameterb4fb3762006-03-14 19:50:20 -080058A. In kernel use of migrate_pages()
59-----------------------------------
Christoph Lametera48d07a2006-02-01 03:05:38 -080060
611. Remove pages from the LRU.
62
63 Lists of pages to be migrated are generated by scanning over
64 pages and moving them into lists. This is done by
Christoph Lameterb4fb3762006-03-14 19:50:20 -080065 calling isolate_lru_page().
Christoph Lametera48d07a2006-02-01 03:05:38 -080066 Calling isolate_lru_page increases the references to the page
Christoph Lameterb4fb3762006-03-14 19:50:20 -080067 so that it cannot vanish while the page migration occurs.
68 It also prevents the swapper or other scans to encounter
69 the page.
Christoph Lametera48d07a2006-02-01 03:05:38 -080070
Christoph Lameter742755a2006-06-23 02:03:55 -0700712. We need to have a function of type new_page_t that can be
72 passed to migrate_pages(). This function should figure out
73 how to allocate the correct new page given the old page.
Christoph Lametera48d07a2006-02-01 03:05:38 -080074
753. The migrate_pages() function is called which attempts
Christoph Lameter742755a2006-06-23 02:03:55 -070076 to do the migration. It will call the function to allocate
77 the new page for each page that is considered for
78 moving.
Christoph Lametera48d07a2006-02-01 03:05:38 -080079
Christoph Lameterb4fb3762006-03-14 19:50:20 -080080B. How migrate_pages() works
81----------------------------
Christoph Lametera48d07a2006-02-01 03:05:38 -080082
Christoph Lameterb4fb3762006-03-14 19:50:20 -080083migrate_pages() does several passes over its list of pages. A page is moved
84if all references to a page are removable at the time. The page has
85already been removed from the LRU via isolate_lru_page() and the refcount
86is increased so that the page cannot be freed while page migration occurs.
Christoph Lametera48d07a2006-02-01 03:05:38 -080087
88Steps:
89
901. Lock the page to be migrated
91
922. Insure that writeback is complete.
93
Christoph Lameter8d3c1382006-06-23 02:03:39 -0700943. Prep the new page that we want to move to. It is locked
Christoph Lametera48d07a2006-02-01 03:05:38 -080095 and set to not being uptodate so that all accesses to the new
Christoph Lameterb4fb3762006-03-14 19:50:20 -080096 page immediately lock while the move is in progress.
Christoph Lametera48d07a2006-02-01 03:05:38 -080097
Christoph Lameter8d3c1382006-06-23 02:03:39 -0700984. The new page is prepped with some settings from the old page so that
99 accesses to the new page will discover a page with the correct settings.
100
1015. All the page table references to the page are converted
102 to migration entries or dropped (nonlinear vmas).
103 This decrease the mapcount of a page. If the resulting
104 mapcount is not zero then we do not migrate the page.
105 All user space processes that attempt to access the page
106 will now wait on the page lock.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800107
Christoph Lameterb4fb3762006-03-14 19:50:20 -08001086. The radix tree lock is taken. This will cause all processes trying
Christoph Lameter8d3c1382006-06-23 02:03:39 -0700109 to access the page via the mapping to block on the radix tree spinlock.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800110
1117. 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
1148. The radix tree is checked and if it does not contain the pointer to this
Christoph Lameter8d3c1382006-06-23 02:03:39 -0700115 page then we back out because someone else modified the radix tree.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800116
Christoph Lameter8d3c1382006-06-23 02:03:39 -07001179. The radix tree is changed to point to the new page.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800118
Christoph Lameter8d3c1382006-06-23 02:03:39 -070011910. The reference count of the old page is dropped because the radix tree
120 reference is gone. A reference to the new page is established because
121 the new page is referenced to by the radix tree.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800122
Christoph Lameter8d3c1382006-06-23 02:03:39 -070012311. The radix tree lock is dropped. With that lookups in the mapping
124 become possible again. Processes will move from spinning on the tree_lock
125 to sleeping on the locked new page.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800126
Christoph Lameter8d3c1382006-06-23 02:03:39 -070012712. The page contents are copied to the new page.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800128
Christoph Lameter8d3c1382006-06-23 02:03:39 -070012913. The remaining page flags are copied to the new page.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800130
Christoph Lameter8d3c1382006-06-23 02:03:39 -070013114. The old page flags are cleared to indicate that the page does
132 not provide any information anymore.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800133
Christoph Lameter8d3c1382006-06-23 02:03:39 -070013415. Queued up writeback on the new page is triggered.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800135
Christoph Lameter8d3c1382006-06-23 02:03:39 -070013616. If migration entries were page then replace them with real ptes. Doing
137 so will enable access for user space processes not already waiting for
138 the page lock.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800139
Christoph Lameterb4fb3762006-03-14 19:50:20 -080014019. The page locks are dropped from the old and new page.
Christoph Lameter8d3c1382006-06-23 02:03:39 -0700141 Processes waiting on the page lock will redo their page faults
142 and will reach the new page.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800143
Christoph Lameterb4fb3762006-03-14 19:50:20 -080014420. The new page is moved to the LRU and can be scanned by the swapper
145 etc again.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800146
Christoph Lameter8d3c1382006-06-23 02:03:39 -0700147Christoph Lameter, May 8, 2006.
Christoph Lametera48d07a2006-02-01 03:05:38 -0800148