1.. _numa_memory_policy:
2
3==================
4NUMA Memory Policy
5==================
6
7What is NUMA Memory Policy?
8============================
9
10In the Linux kernel, "memory policy" determines from which node the kernel will
11allocate memory in a NUMA system or in an emulated NUMA system.  Linux has
12supported platforms with Non-Uniform Memory Access architectures since 2.4.?.
13The current memory policy support was added to Linux 2.6 around May 2004.  This
14document attempts to describe the concepts and APIs of the 2.6 memory policy
15support.
16
17Memory policies should not be confused with cpusets
18(``Documentation/cgroup-v1/cpusets.txt``)
19which is an administrative mechanism for restricting the nodes from which
20memory may be allocated by a set of processes. Memory policies are a
21programming interface that a NUMA-aware application can take advantage of.  When
22both cpusets and policies are applied to a task, the restrictions of the cpuset
23takes priority.  See :ref:`Memory Policies and cpusets <mem_pol_and_cpusets>`
24below for more details.
25
26Memory Policy Concepts
27======================
28
29Scope of Memory Policies
30------------------------
31
32The Linux kernel supports _scopes_ of memory policy, described here from
33most general to most specific:
34
35System Default Policy
36	this policy is "hard coded" into the kernel.  It is the policy
37	that governs all page allocations that aren't controlled by
38	one of the more specific policy scopes discussed below.  When
39	the system is "up and running", the system default policy will
40	use "local allocation" described below.  However, during boot
41	up, the system default policy will be set to interleave
42	allocations across all nodes with "sufficient" memory, so as
43	not to overload the initial boot node with boot-time
44	allocations.
45
46Task/Process Policy
47	this is an optional, per-task policy.  When defined for a
48	specific task, this policy controls all page allocations made
49	by or on behalf of the task that aren't controlled by a more
50	specific scope. If a task does not define a task policy, then
51	all page allocations that would have been controlled by the
52	task policy "fall back" to the System Default Policy.
53
54	The task policy applies to the entire address space of a task. Thus,
55	it is inheritable, and indeed is inherited, across both fork()
56	[clone() w/o the CLONE_VM flag] and exec*().  This allows a parent task
57	to establish the task policy for a child task exec()'d from an
58	executable image that has no awareness of memory policy.  See the
59	:ref:`Memory Policy APIs <memory_policy_apis>` section,
60	below, for an overview of the system call
61	that a task may use to set/change its task/process policy.
62
63	In a multi-threaded task, task policies apply only to the thread
64	[Linux kernel task] that installs the policy and any threads
65	subsequently created by that thread.  Any sibling threads existing
66	at the time a new task policy is installed retain their current
67	policy.
68
69	A task policy applies only to pages allocated after the policy is
70	installed.  Any pages already faulted in by the task when the task
71	changes its task policy remain where they were allocated based on
72	the policy at the time they were allocated.
73
74.. _vma_policy:
75
76VMA Policy
77	A "VMA" or "Virtual Memory Area" refers to a range of a task's
78	virtual address space.  A task may define a specific policy for a range
79	of its virtual address space.   See the
80	:ref:`Memory Policy APIs <memory_policy_apis>` section,
81	below, for an overview of the mbind() system call used to set a VMA
82	policy.
83
84	A VMA policy will govern the allocation of pages that back
85	this region of the address space.  Any regions of the task's
86	address space that don't have an explicit VMA policy will fall
87	back to the task policy, which may itself fall back to the
88	System Default Policy.
89
90	VMA policies have a few complicating details:
91
92	* VMA policy applies ONLY to anonymous pages.  These include
93	  pages allocated for anonymous segments, such as the task
94	  stack and heap, and any regions of the address space
95	  mmap()ed with the MAP_ANONYMOUS flag.  If a VMA policy is
96	  applied to a file mapping, it will be ignored if the mapping
97	  used the MAP_SHARED flag.  If the file mapping used the
98	  MAP_PRIVATE flag, the VMA policy will only be applied when
99	  an anonymous page is allocated on an attempt to write to the
100	  mapping-- i.e., at Copy-On-Write.
101
102	* VMA policies are shared between all tasks that share a
103	  virtual address space--a.k.a. threads--independent of when
104	  the policy is installed; and they are inherited across
105	  fork().  However, because VMA policies refer to a specific
106	  region of a task's address space, and because the address
107	  space is discarded and recreated on exec*(), VMA policies
108	  are NOT inheritable across exec().  Thus, only NUMA-aware
109	  applications may use VMA policies.
110
111	* A task may install a new VMA policy on a sub-range of a
112	  previously mmap()ed region.  When this happens, Linux splits
113	  the existing virtual memory area into 2 or 3 VMAs, each with
114	  it's own policy.
115
116	* By default, VMA policy applies only to pages allocated after
117	  the policy is installed.  Any pages already faulted into the
118	  VMA range remain where they were allocated based on the
119	  policy at the time they were allocated.  However, since
120	  2.6.16, Linux supports page migration via the mbind() system
121	  call, so that page contents can be moved to match a newly
122	  installed policy.
123
124Shared Policy
125	Conceptually, shared policies apply to "memory objects" mapped
126	shared into one or more tasks' distinct address spaces.  An
127	application installs shared policies the same way as VMA
128	policies--using the mbind() system call specifying a range of
129	virtual addresses that map the shared object.  However, unlike
130	VMA policies, which can be considered to be an attribute of a
131	range of a task's address space, shared policies apply
132	directly to the shared object.  Thus, all tasks that attach to
133	the object share the policy, and all pages allocated for the
134	shared object, by any task, will obey the shared policy.
135
136	As of 2.6.22, only shared memory segments, created by shmget() or
137	mmap(MAP_ANONYMOUS|MAP_SHARED), support shared policy.  When shared
138	policy support was added to Linux, the associated data structures were
139	added to hugetlbfs shmem segments.  At the time, hugetlbfs did not
140	support allocation at fault time--a.k.a lazy allocation--so hugetlbfs
141	shmem segments were never "hooked up" to the shared policy support.
142	Although hugetlbfs segments now support lazy allocation, their support
143	for shared policy has not been completed.
144
145	As mentioned above in :ref:`VMA policies <vma_policy>` section,
146	allocations of page cache pages for regular files mmap()ed
147	with MAP_SHARED ignore any VMA policy installed on the virtual
148	address range backed by the shared file mapping.  Rather,
149	shared page cache pages, including pages backing private
150	mappings that have not yet been written by the task, follow
151	task policy, if any, else System Default Policy.
152
153	The shared policy infrastructure supports different policies on subset
154	ranges of the shared object.  However, Linux still splits the VMA of
155	the task that installs the policy for each range of distinct policy.
156	Thus, different tasks that attach to a shared memory segment can have
157	different VMA configurations mapping that one shared object.  This
158	can be seen by examining the /proc/<pid>/numa_maps of tasks sharing
159	a shared memory region, when one task has installed shared policy on
160	one or more ranges of the region.
161
162Components of Memory Policies
163-----------------------------
164
165A NUMA memory policy consists of a "mode", optional mode flags, and
166an optional set of nodes.  The mode determines the behavior of the
167policy, the optional mode flags determine the behavior of the mode,
168and the optional set of nodes can be viewed as the arguments to the
169policy behavior.
170
171Internally, memory policies are implemented by a reference counted
172structure, struct mempolicy.  Details of this structure will be
173discussed in context, below, as required to explain the behavior.
174
175NUMA memory policy supports the following 4 behavioral modes:
176
177Default Mode--MPOL_DEFAULT
178	This mode is only used in the memory policy APIs.  Internally,
179	MPOL_DEFAULT is converted to the NULL memory policy in all
180	policy scopes.  Any existing non-default policy will simply be
181	removed when MPOL_DEFAULT is specified.  As a result,
182	MPOL_DEFAULT means "fall back to the next most specific policy
183	scope."
184
185	For example, a NULL or default task policy will fall back to the
186	system default policy.  A NULL or default vma policy will fall
187	back to the task policy.
188
189	When specified in one of the memory policy APIs, the Default mode
190	does not use the optional set of nodes.
191
192	It is an error for the set of nodes specified for this policy to
193	be non-empty.
194
195MPOL_BIND
196	This mode specifies that memory must come from the set of
197	nodes specified by the policy.  Memory will be allocated from
198	the node in the set with sufficient free memory that is
199	closest to the node where the allocation takes place.
200
201MPOL_PREFERRED
202	This mode specifies that the allocation should be attempted
203	from the single node specified in the policy.  If that
204	allocation fails, the kernel will search other nodes, in order
205	of increasing distance from the preferred node based on
206	information provided by the platform firmware.
207
208	Internally, the Preferred policy uses a single node--the
209	preferred_node member of struct mempolicy.  When the internal
210	mode flag MPOL_F_LOCAL is set, the preferred_node is ignored
211	and the policy is interpreted as local allocation.  "Local"
212	allocation policy can be viewed as a Preferred policy that
213	starts at the node containing the cpu where the allocation
214	takes place.
215
216	It is possible for the user to specify that local allocation
217	is always preferred by passing an empty nodemask with this
218	mode.  If an empty nodemask is passed, the policy cannot use
219	the MPOL_F_STATIC_NODES or MPOL_F_RELATIVE_NODES flags
220	described below.
221
222MPOL_INTERLEAVED
223	This mode specifies that page allocations be interleaved, on a
224	page granularity, across the nodes specified in the policy.
225	This mode also behaves slightly differently, based on the
226	context where it is used:
227
228	For allocation of anonymous pages and shared memory pages,
229	Interleave mode indexes the set of nodes specified by the
230	policy using the page offset of the faulting address into the
231	segment [VMA] containing the address modulo the number of
232	nodes specified by the policy.  It then attempts to allocate a
233	page, starting at the selected node, as if the node had been
234	specified by a Preferred policy or had been selected by a
235	local allocation.  That is, allocation will follow the per
236	node zonelist.
237
238	For allocation of page cache pages, Interleave mode indexes
239	the set of nodes specified by the policy using a node counter
240	maintained per task.  This counter wraps around to the lowest
241	specified node after it reaches the highest specified node.
242	This will tend to spread the pages out over the nodes
243	specified by the policy based on the order in which they are
244	allocated, rather than based on any page offset into an
245	address range or file.  During system boot up, the temporary
246	interleaved system default policy works in this mode.
247
248NUMA memory policy supports the following optional mode flags:
249
250MPOL_F_STATIC_NODES
251	This flag specifies that the nodemask passed by
252	the user should not be remapped if the task or VMA's set of allowed
253	nodes changes after the memory policy has been defined.
254
255	Without this flag, any time a mempolicy is rebound because of a
256	change in the set of allowed nodes, the node (Preferred) or
257	nodemask (Bind, Interleave) is remapped to the new set of
258	allowed nodes.  This may result in nodes being used that were
259	previously undesired.
260
261	With this flag, if the user-specified nodes overlap with the
262	nodes allowed by the task's cpuset, then the memory policy is
263	applied to their intersection.  If the two sets of nodes do not
264	overlap, the Default policy is used.
265
266	For example, consider a task that is attached to a cpuset with
267	mems 1-3 that sets an Interleave policy over the same set.  If
268	the cpuset's mems change to 3-5, the Interleave will now occur
269	over nodes 3, 4, and 5.  With this flag, however, since only node
270	3 is allowed from the user's nodemask, the "interleave" only
271	occurs over that node.  If no nodes from the user's nodemask are
272	now allowed, the Default behavior is used.
273
274	MPOL_F_STATIC_NODES cannot be combined with the
275	MPOL_F_RELATIVE_NODES flag.  It also cannot be used for
276	MPOL_PREFERRED policies that were created with an empty nodemask
277	(local allocation).
278
279MPOL_F_RELATIVE_NODES
280	This flag specifies that the nodemask passed
281	by the user will be mapped relative to the set of the task or VMA's
282	set of allowed nodes.  The kernel stores the user-passed nodemask,
283	and if the allowed nodes changes, then that original nodemask will
284	be remapped relative to the new set of allowed nodes.
285
286	Without this flag (and without MPOL_F_STATIC_NODES), anytime a
287	mempolicy is rebound because of a change in the set of allowed
288	nodes, the node (Preferred) or nodemask (Bind, Interleave) is
289	remapped to the new set of allowed nodes.  That remap may not
290	preserve the relative nature of the user's passed nodemask to its
291	set of allowed nodes upon successive rebinds: a nodemask of
292	1,3,5 may be remapped to 7-9 and then to 1-3 if the set of
293	allowed nodes is restored to its original state.
294
295	With this flag, the remap is done so that the node numbers from
296	the user's passed nodemask are relative to the set of allowed
297	nodes.  In other words, if nodes 0, 2, and 4 are set in the user's
298	nodemask, the policy will be effected over the first (and in the
299	Bind or Interleave case, the third and fifth) nodes in the set of
300	allowed nodes.  The nodemask passed by the user represents nodes
301	relative to task or VMA's set of allowed nodes.
302
303	If the user's nodemask includes nodes that are outside the range
304	of the new set of allowed nodes (for example, node 5 is set in
305	the user's nodemask when the set of allowed nodes is only 0-3),
306	then the remap wraps around to the beginning of the nodemask and,
307	if not already set, sets the node in the mempolicy nodemask.
308
309	For example, consider a task that is attached to a cpuset with
310	mems 2-5 that sets an Interleave policy over the same set with
311	MPOL_F_RELATIVE_NODES.  If the cpuset's mems change to 3-7, the
312	interleave now occurs over nodes 3,5-7.  If the cpuset's mems
313	then change to 0,2-3,5, then the interleave occurs over nodes
314	0,2-3,5.
315
316	Thanks to the consistent remapping, applications preparing
317	nodemasks to specify memory policies using this flag should
318	disregard their current, actual cpuset imposed memory placement
319	and prepare the nodemask as if they were always located on
320	memory nodes 0 to N-1, where N is the number of memory nodes the
321	policy is intended to manage.  Let the kernel then remap to the
322	set of memory nodes allowed by the task's cpuset, as that may
323	change over time.
324
325	MPOL_F_RELATIVE_NODES cannot be combined with the
326	MPOL_F_STATIC_NODES flag.  It also cannot be used for
327	MPOL_PREFERRED policies that were created with an empty nodemask
328	(local allocation).
329
330Memory Policy Reference Counting
331================================
332
333To resolve use/free races, struct mempolicy contains an atomic reference
334count field.  Internal interfaces, mpol_get()/mpol_put() increment and
335decrement this reference count, respectively.  mpol_put() will only free
336the structure back to the mempolicy kmem cache when the reference count
337goes to zero.
338
339When a new memory policy is allocated, its reference count is initialized
340to '1', representing the reference held by the task that is installing the
341new policy.  When a pointer to a memory policy structure is stored in another
342structure, another reference is added, as the task's reference will be dropped
343on completion of the policy installation.
344
345During run-time "usage" of the policy, we attempt to minimize atomic operations
346on the reference count, as this can lead to cache lines bouncing between cpus
347and NUMA nodes.  "Usage" here means one of the following:
348
3491) querying of the policy, either by the task itself [using the get_mempolicy()
350   API discussed below] or by another task using the /proc/<pid>/numa_maps
351   interface.
352
3532) examination of the policy to determine the policy mode and associated node
354   or node lists, if any, for page allocation.  This is considered a "hot
355   path".  Note that for MPOL_BIND, the "usage" extends across the entire
356   allocation process, which may sleep during page reclaimation, because the
357   BIND policy nodemask is used, by reference, to filter ineligible nodes.
358
359We can avoid taking an extra reference during the usages listed above as
360follows:
361
3621) we never need to get/free the system default policy as this is never
363   changed nor freed, once the system is up and running.
364
3652) for querying the policy, we do not need to take an extra reference on the
366   target task's task policy nor vma policies because we always acquire the
367   task's mm's mmap_sem for read during the query.  The set_mempolicy() and
368   mbind() APIs [see below] always acquire the mmap_sem for write when
369   installing or replacing task or vma policies.  Thus, there is no possibility
370   of a task or thread freeing a policy while another task or thread is
371   querying it.
372
3733) Page allocation usage of task or vma policy occurs in the fault path where
374   we hold them mmap_sem for read.  Again, because replacing the task or vma
375   policy requires that the mmap_sem be held for write, the policy can't be
376   freed out from under us while we're using it for page allocation.
377
3784) Shared policies require special consideration.  One task can replace a
379   shared memory policy while another task, with a distinct mmap_sem, is
380   querying or allocating a page based on the policy.  To resolve this
381   potential race, the shared policy infrastructure adds an extra reference
382   to the shared policy during lookup while holding a spin lock on the shared
383   policy management structure.  This requires that we drop this extra
384   reference when we're finished "using" the policy.  We must drop the
385   extra reference on shared policies in the same query/allocation paths
386   used for non-shared policies.  For this reason, shared policies are marked
387   as such, and the extra reference is dropped "conditionally"--i.e., only
388   for shared policies.
389
390   Because of this extra reference counting, and because we must lookup
391   shared policies in a tree structure under spinlock, shared policies are
392   more expensive to use in the page allocation path.  This is especially
393   true for shared policies on shared memory regions shared by tasks running
394   on different NUMA nodes.  This extra overhead can be avoided by always
395   falling back to task or system default policy for shared memory regions,
396   or by prefaulting the entire shared memory region into memory and locking
397   it down.  However, this might not be appropriate for all applications.
398
399.. _memory_policy_apis:
400
401Memory Policy APIs
402==================
403
404Linux supports 3 system calls for controlling memory policy.  These APIS
405always affect only the calling task, the calling task's address space, or
406some shared object mapped into the calling task's address space.
407
408.. note::
409   the headers that define these APIs and the parameter data types for
410   user space applications reside in a package that is not part of the
411   Linux kernel.  The kernel system call interfaces, with the 'sys\_'
412   prefix, are defined in <linux/syscalls.h>; the mode and flag
413   definitions are defined in <linux/mempolicy.h>.
414
415Set [Task] Memory Policy::
416
417	long set_mempolicy(int mode, const unsigned long *nmask,
418					unsigned long maxnode);
419
420Set's the calling task's "task/process memory policy" to mode
421specified by the 'mode' argument and the set of nodes defined by
422'nmask'.  'nmask' points to a bit mask of node ids containing at least
423'maxnode' ids.  Optional mode flags may be passed by combining the
424'mode' argument with the flag (for example: MPOL_INTERLEAVE |
425MPOL_F_STATIC_NODES).
426
427See the set_mempolicy(2) man page for more details
428
429
430Get [Task] Memory Policy or Related Information::
431
432	long get_mempolicy(int *mode,
433			   const unsigned long *nmask, unsigned long maxnode,
434			   void *addr, int flags);
435
436Queries the "task/process memory policy" of the calling task, or the
437policy or location of a specified virtual address, depending on the
438'flags' argument.
439
440See the get_mempolicy(2) man page for more details
441
442
443Install VMA/Shared Policy for a Range of Task's Address Space::
444
445	long mbind(void *start, unsigned long len, int mode,
446		   const unsigned long *nmask, unsigned long maxnode,
447		   unsigned flags);
448
449mbind() installs the policy specified by (mode, nmask, maxnodes) as a
450VMA policy for the range of the calling task's address space specified
451by the 'start' and 'len' arguments.  Additional actions may be
452requested via the 'flags' argument.
453
454See the mbind(2) man page for more details.
455
456Memory Policy Command Line Interface
457====================================
458
459Although not strictly part of the Linux implementation of memory policy,
460a command line tool, numactl(8), exists that allows one to:
461
462+ set the task policy for a specified program via set_mempolicy(2), fork(2) and
463  exec(2)
464
465+ set the shared policy for a shared memory segment via mbind(2)
466
467The numactl(8) tool is packaged with the run-time version of the library
468containing the memory policy system call wrappers.  Some distributions
469package the headers and compile-time libraries in a separate development
470package.
471
472.. _mem_pol_and_cpusets:
473
474Memory Policies and cpusets
475===========================
476
477Memory policies work within cpusets as described above.  For memory policies
478that require a node or set of nodes, the nodes are restricted to the set of
479nodes whose memories are allowed by the cpuset constraints.  If the nodemask
480specified for the policy contains nodes that are not allowed by the cpuset and
481MPOL_F_RELATIVE_NODES is not used, the intersection of the set of nodes
482specified for the policy and the set of nodes with memory is used.  If the
483result is the empty set, the policy is considered invalid and cannot be
484installed.  If MPOL_F_RELATIVE_NODES is used, the policy's nodes are mapped
485onto and folded into the task's set of allowed nodes as previously described.
486
487The interaction of memory policies and cpusets can be problematic when tasks
488in two cpusets share access to a memory region, such as shared memory segments
489created by shmget() of mmap() with the MAP_ANONYMOUS and MAP_SHARED flags, and
490any of the tasks install shared policy on the region, only nodes whose
491memories are allowed in both cpusets may be used in the policies.  Obtaining
492this information requires "stepping outside" the memory policy APIs to use the
493cpuset information and requires that one know in what cpusets other task might
494be attaching to the shared region.  Furthermore, if the cpusets' allowed
495memory sets are disjoint, "local" allocation is the only valid policy.
496