xref: /linux/mm/Kconfig (revision c4101e55974cc7d835fbd2d8e01553a3f61e9e75)
1# SPDX-License-Identifier: GPL-2.0-only
2
3menu "Memory Management options"
4
5#
6# For some reason microblaze and nios2 hard code SWAP=n.  Hopefully we can
7# add proper SWAP support to them, in which case this can be remove.
8#
9config ARCH_NO_SWAP
10	bool
11
12config ZPOOL
13	bool
14
15menuconfig SWAP
16	bool "Support for paging of anonymous memory (swap)"
17	depends on MMU && BLOCK && !ARCH_NO_SWAP
18	default y
19	help
20	  This option allows you to choose whether you want to have support
21	  for so called swap devices or swap files in your kernel that are
22	  used to provide more virtual memory than the actual RAM present
23	  in your computer.  If unsure say Y.
24
25config ZSWAP
26	bool "Compressed cache for swap pages"
27	depends on SWAP
28	select CRYPTO
29	select ZPOOL
30	help
31	  A lightweight compressed cache for swap pages.  It takes
32	  pages that are in the process of being swapped out and attempts to
33	  compress them into a dynamically allocated RAM-based memory pool.
34	  This can result in a significant I/O reduction on swap device and,
35	  in the case where decompressing from RAM is faster than swap device
36	  reads, can also improve workload performance.
37
38config ZSWAP_DEFAULT_ON
39	bool "Enable the compressed cache for swap pages by default"
40	depends on ZSWAP
41	help
42	  If selected, the compressed cache for swap pages will be enabled
43	  at boot, otherwise it will be disabled.
44
45	  The selection made here can be overridden by using the kernel
46	  command line 'zswap.enabled=' option.
47
48config ZSWAP_EXCLUSIVE_LOADS_DEFAULT_ON
49	bool "Invalidate zswap entries when pages are loaded"
50	depends on ZSWAP
51	help
52	  If selected, exclusive loads for zswap will be enabled at boot,
53	  otherwise it will be disabled.
54
55	  If exclusive loads are enabled, when a page is loaded from zswap,
56	  the zswap entry is invalidated at once, as opposed to leaving it
57	  in zswap until the swap entry is freed.
58
59	  This avoids having two copies of the same page in memory
60	  (compressed and uncompressed) after faulting in a page from zswap.
61	  The cost is that if the page was never dirtied and needs to be
62	  swapped out again, it will be re-compressed.
63
64config ZSWAP_SHRINKER_DEFAULT_ON
65	bool "Shrink the zswap pool on memory pressure"
66	depends on ZSWAP
67	default n
68	help
69	  If selected, the zswap shrinker will be enabled, and the pages
70	  stored in the zswap pool will become available for reclaim (i.e
71	  written back to the backing swap device) on memory pressure.
72
73	  This means that zswap writeback could happen even if the pool is
74	  not yet full, or the cgroup zswap limit has not been reached,
75	  reducing the chance that cold pages will reside in the zswap pool
76	  and consume memory indefinitely.
77
78choice
79	prompt "Default compressor"
80	depends on ZSWAP
81	default ZSWAP_COMPRESSOR_DEFAULT_LZO
82	help
83	  Selects the default compression algorithm for the compressed cache
84	  for swap pages.
85
86	  For an overview what kind of performance can be expected from
87	  a particular compression algorithm please refer to the benchmarks
88	  available at the following LWN page:
89	  https://lwn.net/Articles/751795/
90
91	  If in doubt, select 'LZO'.
92
93	  The selection made here can be overridden by using the kernel
94	  command line 'zswap.compressor=' option.
95
96config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
97	bool "Deflate"
98	select CRYPTO_DEFLATE
99	help
100	  Use the Deflate algorithm as the default compression algorithm.
101
102config ZSWAP_COMPRESSOR_DEFAULT_LZO
103	bool "LZO"
104	select CRYPTO_LZO
105	help
106	  Use the LZO algorithm as the default compression algorithm.
107
108config ZSWAP_COMPRESSOR_DEFAULT_842
109	bool "842"
110	select CRYPTO_842
111	help
112	  Use the 842 algorithm as the default compression algorithm.
113
114config ZSWAP_COMPRESSOR_DEFAULT_LZ4
115	bool "LZ4"
116	select CRYPTO_LZ4
117	help
118	  Use the LZ4 algorithm as the default compression algorithm.
119
120config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
121	bool "LZ4HC"
122	select CRYPTO_LZ4HC
123	help
124	  Use the LZ4HC algorithm as the default compression algorithm.
125
126config ZSWAP_COMPRESSOR_DEFAULT_ZSTD
127	bool "zstd"
128	select CRYPTO_ZSTD
129	help
130	  Use the zstd algorithm as the default compression algorithm.
131endchoice
132
133config ZSWAP_COMPRESSOR_DEFAULT
134       string
135       depends on ZSWAP
136       default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
137       default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO
138       default "842" if ZSWAP_COMPRESSOR_DEFAULT_842
139       default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4
140       default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
141       default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD
142       default ""
143
144choice
145	prompt "Default allocator"
146	depends on ZSWAP
147	default ZSWAP_ZPOOL_DEFAULT_ZSMALLOC if MMU
148	default ZSWAP_ZPOOL_DEFAULT_ZBUD
149	help
150	  Selects the default allocator for the compressed cache for
151	  swap pages.
152	  The default is 'zbud' for compatibility, however please do
153	  read the description of each of the allocators below before
154	  making a right choice.
155
156	  The selection made here can be overridden by using the kernel
157	  command line 'zswap.zpool=' option.
158
159config ZSWAP_ZPOOL_DEFAULT_ZBUD
160	bool "zbud"
161	select ZBUD
162	help
163	  Use the zbud allocator as the default allocator.
164
165config ZSWAP_ZPOOL_DEFAULT_Z3FOLD
166	bool "z3fold"
167	select Z3FOLD
168	help
169	  Use the z3fold allocator as the default allocator.
170
171config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
172	bool "zsmalloc"
173	select ZSMALLOC
174	help
175	  Use the zsmalloc allocator as the default allocator.
176endchoice
177
178config ZSWAP_ZPOOL_DEFAULT
179       string
180       depends on ZSWAP
181       default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD
182       default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD
183       default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
184       default ""
185
186config ZBUD
187	tristate "2:1 compression allocator (zbud)"
188	depends on ZSWAP
189	help
190	  A special purpose allocator for storing compressed pages.
191	  It is designed to store up to two compressed pages per physical
192	  page.  While this design limits storage density, it has simple and
193	  deterministic reclaim properties that make it preferable to a higher
194	  density approach when reclaim will be used.
195
196config Z3FOLD
197	tristate "3:1 compression allocator (z3fold)"
198	depends on ZSWAP
199	help
200	  A special purpose allocator for storing compressed pages.
201	  It is designed to store up to three compressed pages per physical
202	  page. It is a ZBUD derivative so the simplicity and determinism are
203	  still there.
204
205config ZSMALLOC
206	tristate
207	prompt "N:1 compression allocator (zsmalloc)" if ZSWAP
208	depends on MMU
209	help
210	  zsmalloc is a slab-based memory allocator designed to store
211	  pages of various compression levels efficiently. It achieves
212	  the highest storage density with the least amount of fragmentation.
213
214config ZSMALLOC_STAT
215	bool "Export zsmalloc statistics"
216	depends on ZSMALLOC
217	select DEBUG_FS
218	help
219	  This option enables code in the zsmalloc to collect various
220	  statistics about what's happening in zsmalloc and exports that
221	  information to userspace via debugfs.
222	  If unsure, say N.
223
224config ZSMALLOC_CHAIN_SIZE
225	int "Maximum number of physical pages per-zspage"
226	default 8
227	range 4 16
228	depends on ZSMALLOC
229	help
230	  This option sets the upper limit on the number of physical pages
231	  that a zmalloc page (zspage) can consist of. The optimal zspage
232	  chain size is calculated for each size class during the
233	  initialization of the pool.
234
235	  Changing this option can alter the characteristics of size classes,
236	  such as the number of pages per zspage and the number of objects
237	  per zspage. This can also result in different configurations of
238	  the pool, as zsmalloc merges size classes with similar
239	  characteristics.
240
241	  For more information, see zsmalloc documentation.
242
243menu "Slab allocator options"
244
245config SLUB
246	def_bool y
247
248config SLUB_TINY
249	bool "Configure for minimal memory footprint"
250	depends on EXPERT
251	select SLAB_MERGE_DEFAULT
252	help
253	   Configures the slab allocator in a way to achieve minimal memory
254	   footprint, sacrificing scalability, debugging and other features.
255	   This is intended only for the smallest system that had used the
256	   SLOB allocator and is not recommended for systems with more than
257	   16MB RAM.
258
259	   If unsure, say N.
260
261config SLAB_MERGE_DEFAULT
262	bool "Allow slab caches to be merged"
263	default y
264	help
265	  For reduced kernel memory fragmentation, slab caches can be
266	  merged when they share the same size and other characteristics.
267	  This carries a risk of kernel heap overflows being able to
268	  overwrite objects from merged caches (and more easily control
269	  cache layout), which makes such heap attacks easier to exploit
270	  by attackers. By keeping caches unmerged, these kinds of exploits
271	  can usually only damage objects in the same cache. To disable
272	  merging at runtime, "slab_nomerge" can be passed on the kernel
273	  command line.
274
275config SLAB_FREELIST_RANDOM
276	bool "Randomize slab freelist"
277	depends on !SLUB_TINY
278	help
279	  Randomizes the freelist order used on creating new pages. This
280	  security feature reduces the predictability of the kernel slab
281	  allocator against heap overflows.
282
283config SLAB_FREELIST_HARDENED
284	bool "Harden slab freelist metadata"
285	depends on !SLUB_TINY
286	help
287	  Many kernel heap attacks try to target slab cache metadata and
288	  other infrastructure. This options makes minor performance
289	  sacrifices to harden the kernel slab allocator against common
290	  freelist exploit methods.
291
292config SLUB_STATS
293	default n
294	bool "Enable performance statistics"
295	depends on SYSFS && !SLUB_TINY
296	help
297	  The statistics are useful to debug slab allocation behavior in
298	  order find ways to optimize the allocator. This should never be
299	  enabled for production use since keeping statistics slows down
300	  the allocator by a few percentage points. The slabinfo command
301	  supports the determination of the most active slabs to figure
302	  out which slabs are relevant to a particular load.
303	  Try running: slabinfo -DA
304
305config SLUB_CPU_PARTIAL
306	default y
307	depends on SMP && !SLUB_TINY
308	bool "Enable per cpu partial caches"
309	help
310	  Per cpu partial caches accelerate objects allocation and freeing
311	  that is local to a processor at the price of more indeterminism
312	  in the latency of the free. On overflow these caches will be cleared
313	  which requires the taking of locks that may cause latency spikes.
314	  Typically one would choose no for a realtime system.
315
316config RANDOM_KMALLOC_CACHES
317	default n
318	depends on !SLUB_TINY
319	bool "Randomize slab caches for normal kmalloc"
320	help
321	  A hardening feature that creates multiple copies of slab caches for
322	  normal kmalloc allocation and makes kmalloc randomly pick one based
323	  on code address, which makes the attackers more difficult to spray
324	  vulnerable memory objects on the heap for the purpose of exploiting
325	  memory vulnerabilities.
326
327	  Currently the number of copies is set to 16, a reasonably large value
328	  that effectively diverges the memory objects allocated for different
329	  subsystems or modules into different caches, at the expense of a
330	  limited degree of memory and CPU overhead that relates to hardware and
331	  system workload.
332
333endmenu # Slab allocator options
334
335config SHUFFLE_PAGE_ALLOCATOR
336	bool "Page allocator randomization"
337	default SLAB_FREELIST_RANDOM && ACPI_NUMA
338	help
339	  Randomization of the page allocator improves the average
340	  utilization of a direct-mapped memory-side-cache. See section
341	  5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
342	  6.2a specification for an example of how a platform advertises
343	  the presence of a memory-side-cache. There are also incidental
344	  security benefits as it reduces the predictability of page
345	  allocations to compliment SLAB_FREELIST_RANDOM, but the
346	  default granularity of shuffling on the MAX_PAGE_ORDER i.e, 10th
347	  order of pages is selected based on cache utilization benefits
348	  on x86.
349
350	  While the randomization improves cache utilization it may
351	  negatively impact workloads on platforms without a cache. For
352	  this reason, by default, the randomization is enabled only
353	  after runtime detection of a direct-mapped memory-side-cache.
354	  Otherwise, the randomization may be force enabled with the
355	  'page_alloc.shuffle' kernel command line parameter.
356
357	  Say Y if unsure.
358
359config COMPAT_BRK
360	bool "Disable heap randomization"
361	default y
362	help
363	  Randomizing heap placement makes heap exploits harder, but it
364	  also breaks ancient binaries (including anything libc5 based).
365	  This option changes the bootup default to heap randomization
366	  disabled, and can be overridden at runtime by setting
367	  /proc/sys/kernel/randomize_va_space to 2.
368
369	  On non-ancient distros (post-2000 ones) N is usually a safe choice.
370
371config MMAP_ALLOW_UNINITIALIZED
372	bool "Allow mmapped anonymous memory to be uninitialized"
373	depends on EXPERT && !MMU
374	default n
375	help
376	  Normally, and according to the Linux spec, anonymous memory obtained
377	  from mmap() has its contents cleared before it is passed to
378	  userspace.  Enabling this config option allows you to request that
379	  mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
380	  providing a huge performance boost.  If this option is not enabled,
381	  then the flag will be ignored.
382
383	  This is taken advantage of by uClibc's malloc(), and also by
384	  ELF-FDPIC binfmt's brk and stack allocator.
385
386	  Because of the obvious security issues, this option should only be
387	  enabled on embedded devices where you control what is run in
388	  userspace.  Since that isn't generally a problem on no-MMU systems,
389	  it is normally safe to say Y here.
390
391	  See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
392
393config SELECT_MEMORY_MODEL
394	def_bool y
395	depends on ARCH_SELECT_MEMORY_MODEL
396
397choice
398	prompt "Memory model"
399	depends on SELECT_MEMORY_MODEL
400	default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT
401	default FLATMEM_MANUAL
402	help
403	  This option allows you to change some of the ways that
404	  Linux manages its memory internally. Most users will
405	  only have one option here selected by the architecture
406	  configuration. This is normal.
407
408config FLATMEM_MANUAL
409	bool "Flat Memory"
410	depends on !ARCH_SPARSEMEM_ENABLE || ARCH_FLATMEM_ENABLE
411	help
412	  This option is best suited for non-NUMA systems with
413	  flat address space. The FLATMEM is the most efficient
414	  system in terms of performance and resource consumption
415	  and it is the best option for smaller systems.
416
417	  For systems that have holes in their physical address
418	  spaces and for features like NUMA and memory hotplug,
419	  choose "Sparse Memory".
420
421	  If unsure, choose this option (Flat Memory) over any other.
422
423config SPARSEMEM_MANUAL
424	bool "Sparse Memory"
425	depends on ARCH_SPARSEMEM_ENABLE
426	help
427	  This will be the only option for some systems, including
428	  memory hot-plug systems.  This is normal.
429
430	  This option provides efficient support for systems with
431	  holes is their physical address space and allows memory
432	  hot-plug and hot-remove.
433
434	  If unsure, choose "Flat Memory" over this option.
435
436endchoice
437
438config SPARSEMEM
439	def_bool y
440	depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL
441
442config FLATMEM
443	def_bool y
444	depends on !SPARSEMEM || FLATMEM_MANUAL
445
446#
447# SPARSEMEM_EXTREME (which is the default) does some bootmem
448# allocations when sparse_init() is called.  If this cannot
449# be done on your architecture, select this option.  However,
450# statically allocating the mem_section[] array can potentially
451# consume vast quantities of .bss, so be careful.
452#
453# This option will also potentially produce smaller runtime code
454# with gcc 3.4 and later.
455#
456config SPARSEMEM_STATIC
457	bool
458
459#
460# Architecture platforms which require a two level mem_section in SPARSEMEM
461# must select this option. This is usually for architecture platforms with
462# an extremely sparse physical address space.
463#
464config SPARSEMEM_EXTREME
465	def_bool y
466	depends on SPARSEMEM && !SPARSEMEM_STATIC
467
468config SPARSEMEM_VMEMMAP_ENABLE
469	bool
470
471config SPARSEMEM_VMEMMAP
472	bool "Sparse Memory virtual memmap"
473	depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE
474	default y
475	help
476	  SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise
477	  pfn_to_page and page_to_pfn operations.  This is the most
478	  efficient option when sufficient kernel resources are available.
479#
480# Select this config option from the architecture Kconfig, if it is preferred
481# to enable the feature of HugeTLB/dev_dax vmemmap optimization.
482#
483config ARCH_WANT_OPTIMIZE_DAX_VMEMMAP
484	bool
485
486config ARCH_WANT_OPTIMIZE_HUGETLB_VMEMMAP
487	bool
488
489config HAVE_MEMBLOCK_PHYS_MAP
490	bool
491
492config HAVE_FAST_GUP
493	depends on MMU
494	bool
495
496# Don't discard allocated memory used to track "memory" and "reserved" memblocks
497# after early boot, so it can still be used to test for validity of memory.
498# Also, memblocks are updated with memory hot(un)plug.
499config ARCH_KEEP_MEMBLOCK
500	bool
501
502# Keep arch NUMA mapping infrastructure post-init.
503config NUMA_KEEP_MEMINFO
504	bool
505
506config MEMORY_ISOLATION
507	bool
508
509# IORESOURCE_SYSTEM_RAM regions in the kernel resource tree that are marked
510# IORESOURCE_EXCLUSIVE cannot be mapped to user space, for example, via
511# /dev/mem.
512config EXCLUSIVE_SYSTEM_RAM
513	def_bool y
514	depends on !DEVMEM || STRICT_DEVMEM
515
516#
517# Only be set on architectures that have completely implemented memory hotplug
518# feature. If you are not sure, don't touch it.
519#
520config HAVE_BOOTMEM_INFO_NODE
521	def_bool n
522
523config ARCH_ENABLE_MEMORY_HOTPLUG
524	bool
525
526config ARCH_ENABLE_MEMORY_HOTREMOVE
527	bool
528
529# eventually, we can have this option just 'select SPARSEMEM'
530menuconfig MEMORY_HOTPLUG
531	bool "Memory hotplug"
532	select MEMORY_ISOLATION
533	depends on SPARSEMEM
534	depends on ARCH_ENABLE_MEMORY_HOTPLUG
535	depends on 64BIT
536	select NUMA_KEEP_MEMINFO if NUMA
537
538if MEMORY_HOTPLUG
539
540config MEMORY_HOTPLUG_DEFAULT_ONLINE
541	bool "Online the newly added memory blocks by default"
542	depends on MEMORY_HOTPLUG
543	help
544	  This option sets the default policy setting for memory hotplug
545	  onlining policy (/sys/devices/system/memory/auto_online_blocks) which
546	  determines what happens to newly added memory regions. Policy setting
547	  can always be changed at runtime.
548	  See Documentation/admin-guide/mm/memory-hotplug.rst for more information.
549
550	  Say Y here if you want all hot-plugged memory blocks to appear in
551	  'online' state by default.
552	  Say N here if you want the default policy to keep all hot-plugged
553	  memory blocks in 'offline' state.
554
555config MEMORY_HOTREMOVE
556	bool "Allow for memory hot remove"
557	select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64)
558	depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
559	depends on MIGRATION
560
561config MHP_MEMMAP_ON_MEMORY
562	def_bool y
563	depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP
564	depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
565
566endif # MEMORY_HOTPLUG
567
568config ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
569       bool
570
571# Heavily threaded applications may benefit from splitting the mm-wide
572# page_table_lock, so that faults on different parts of the user address
573# space can be handled with less contention: split it at this NR_CPUS.
574# Default to 4 for wider testing, though 8 might be more appropriate.
575# ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
576# PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
577# SPARC32 allocates multiple pte tables within a single page, and therefore
578# a per-page lock leads to problems when multiple tables need to be locked
579# at the same time (e.g. copy_page_range()).
580# DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page.
581#
582config SPLIT_PTLOCK_CPUS
583	int
584	default "999999" if !MMU
585	default "999999" if ARM && !CPU_CACHE_VIPT
586	default "999999" if PARISC && !PA20
587	default "999999" if SPARC32
588	default "4"
589
590config ARCH_ENABLE_SPLIT_PMD_PTLOCK
591	bool
592
593#
594# support for memory balloon
595config MEMORY_BALLOON
596	bool
597
598#
599# support for memory balloon compaction
600config BALLOON_COMPACTION
601	bool "Allow for balloon memory compaction/migration"
602	def_bool y
603	depends on COMPACTION && MEMORY_BALLOON
604	help
605	  Memory fragmentation introduced by ballooning might reduce
606	  significantly the number of 2MB contiguous memory blocks that can be
607	  used within a guest, thus imposing performance penalties associated
608	  with the reduced number of transparent huge pages that could be used
609	  by the guest workload. Allowing the compaction & migration for memory
610	  pages enlisted as being part of memory balloon devices avoids the
611	  scenario aforementioned and helps improving memory defragmentation.
612
613#
614# support for memory compaction
615config COMPACTION
616	bool "Allow for memory compaction"
617	def_bool y
618	select MIGRATION
619	depends on MMU
620	help
621	  Compaction is the only memory management component to form
622	  high order (larger physically contiguous) memory blocks
623	  reliably. The page allocator relies on compaction heavily and
624	  the lack of the feature can lead to unexpected OOM killer
625	  invocations for high order memory requests. You shouldn't
626	  disable this option unless there really is a strong reason for
627	  it and then we would be really interested to hear about that at
628	  linux-mm@kvack.org.
629
630config COMPACT_UNEVICTABLE_DEFAULT
631	int
632	depends on COMPACTION
633	default 0 if PREEMPT_RT
634	default 1
635
636#
637# support for free page reporting
638config PAGE_REPORTING
639	bool "Free page reporting"
640	def_bool n
641	help
642	  Free page reporting allows for the incremental acquisition of
643	  free pages from the buddy allocator for the purpose of reporting
644	  those pages to another entity, such as a hypervisor, so that the
645	  memory can be freed within the host for other uses.
646
647#
648# support for page migration
649#
650config MIGRATION
651	bool "Page migration"
652	def_bool y
653	depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU
654	help
655	  Allows the migration of the physical location of pages of processes
656	  while the virtual addresses are not changed. This is useful in
657	  two situations. The first is on NUMA systems to put pages nearer
658	  to the processors accessing. The second is when allocating huge
659	  pages as migration can relocate pages to satisfy a huge page
660	  allocation instead of reclaiming.
661
662config DEVICE_MIGRATION
663	def_bool MIGRATION && ZONE_DEVICE
664
665config ARCH_ENABLE_HUGEPAGE_MIGRATION
666	bool
667
668config ARCH_ENABLE_THP_MIGRATION
669	bool
670
671config HUGETLB_PAGE_SIZE_VARIABLE
672	def_bool n
673	help
674	  Allows the pageblock_order value to be dynamic instead of just standard
675	  HUGETLB_PAGE_ORDER when there are multiple HugeTLB page sizes available
676	  on a platform.
677
678	  Note that the pageblock_order cannot exceed MAX_PAGE_ORDER and will be
679	  clamped down to MAX_PAGE_ORDER.
680
681config CONTIG_ALLOC
682	def_bool (MEMORY_ISOLATION && COMPACTION) || CMA
683
684config PCP_BATCH_SCALE_MAX
685	int "Maximum scale factor of PCP (Per-CPU pageset) batch allocate/free"
686	default 5
687	range 0 6
688	help
689	  In page allocator, PCP (Per-CPU pageset) is refilled and drained in
690	  batches.  The batch number is scaled automatically to improve page
691	  allocation/free throughput.  But too large scale factor may hurt
692	  latency.  This option sets the upper limit of scale factor to limit
693	  the maximum latency.
694
695config PHYS_ADDR_T_64BIT
696	def_bool 64BIT
697
698config BOUNCE
699	bool "Enable bounce buffers"
700	default y
701	depends on BLOCK && MMU && HIGHMEM
702	help
703	  Enable bounce buffers for devices that cannot access the full range of
704	  memory available to the CPU. Enabled by default when HIGHMEM is
705	  selected, but you may say n to override this.
706
707config MMU_NOTIFIER
708	bool
709	select INTERVAL_TREE
710
711config KSM
712	bool "Enable KSM for page merging"
713	depends on MMU
714	select XXHASH
715	help
716	  Enable Kernel Samepage Merging: KSM periodically scans those areas
717	  of an application's address space that an app has advised may be
718	  mergeable.  When it finds pages of identical content, it replaces
719	  the many instances by a single page with that content, so
720	  saving memory until one or another app needs to modify the content.
721	  Recommended for use with KVM, or with other duplicative applications.
722	  See Documentation/mm/ksm.rst for more information: KSM is inactive
723	  until a program has madvised that an area is MADV_MERGEABLE, and
724	  root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
725
726config DEFAULT_MMAP_MIN_ADDR
727	int "Low address space to protect from user allocation"
728	depends on MMU
729	default 4096
730	help
731	  This is the portion of low virtual memory which should be protected
732	  from userspace allocation.  Keeping a user from writing to low pages
733	  can help reduce the impact of kernel NULL pointer bugs.
734
735	  For most ppc64 and x86 users with lots of address space
736	  a value of 65536 is reasonable and should cause no problems.
737	  On arm and other archs it should not be higher than 32768.
738	  Programs which use vm86 functionality or have some need to map
739	  this low address space will need CAP_SYS_RAWIO or disable this
740	  protection by setting the value to 0.
741
742	  This value can be changed after boot using the
743	  /proc/sys/vm/mmap_min_addr tunable.
744
745config ARCH_SUPPORTS_MEMORY_FAILURE
746	bool
747
748config MEMORY_FAILURE
749	depends on MMU
750	depends on ARCH_SUPPORTS_MEMORY_FAILURE
751	bool "Enable recovery from hardware memory errors"
752	select MEMORY_ISOLATION
753	select RAS
754	help
755	  Enables code to recover from some memory failures on systems
756	  with MCA recovery. This allows a system to continue running
757	  even when some of its memory has uncorrected errors. This requires
758	  special hardware support and typically ECC memory.
759
760config HWPOISON_INJECT
761	tristate "HWPoison pages injector"
762	depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS
763	select PROC_PAGE_MONITOR
764
765config NOMMU_INITIAL_TRIM_EXCESS
766	int "Turn on mmap() excess space trimming before booting"
767	depends on !MMU
768	default 1
769	help
770	  The NOMMU mmap() frequently needs to allocate large contiguous chunks
771	  of memory on which to store mappings, but it can only ask the system
772	  allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
773	  more than it requires.  To deal with this, mmap() is able to trim off
774	  the excess and return it to the allocator.
775
776	  If trimming is enabled, the excess is trimmed off and returned to the
777	  system allocator, which can cause extra fragmentation, particularly
778	  if there are a lot of transient processes.
779
780	  If trimming is disabled, the excess is kept, but not used, which for
781	  long-term mappings means that the space is wasted.
782
783	  Trimming can be dynamically controlled through a sysctl option
784	  (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of
785	  excess pages there must be before trimming should occur, or zero if
786	  no trimming is to occur.
787
788	  This option specifies the initial value of this option.  The default
789	  of 1 says that all excess pages should be trimmed.
790
791	  See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
792
793config ARCH_WANT_GENERAL_HUGETLB
794	bool
795
796config ARCH_WANTS_THP_SWAP
797	def_bool n
798
799menuconfig TRANSPARENT_HUGEPAGE
800	bool "Transparent Hugepage Support"
801	depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE && !PREEMPT_RT
802	select COMPACTION
803	select XARRAY_MULTI
804	help
805	  Transparent Hugepages allows the kernel to use huge pages and
806	  huge tlb transparently to the applications whenever possible.
807	  This feature can improve computing performance to certain
808	  applications by speeding up page faults during memory
809	  allocation, by reducing the number of tlb misses and by speeding
810	  up the pagetable walking.
811
812	  If memory constrained on embedded, you may want to say N.
813
814if TRANSPARENT_HUGEPAGE
815
816choice
817	prompt "Transparent Hugepage Support sysfs defaults"
818	depends on TRANSPARENT_HUGEPAGE
819	default TRANSPARENT_HUGEPAGE_ALWAYS
820	help
821	  Selects the sysfs defaults for Transparent Hugepage Support.
822
823	config TRANSPARENT_HUGEPAGE_ALWAYS
824		bool "always"
825	help
826	  Enabling Transparent Hugepage always, can increase the
827	  memory footprint of applications without a guaranteed
828	  benefit but it will work automatically for all applications.
829
830	config TRANSPARENT_HUGEPAGE_MADVISE
831		bool "madvise"
832	help
833	  Enabling Transparent Hugepage madvise, will only provide a
834	  performance improvement benefit to the applications using
835	  madvise(MADV_HUGEPAGE) but it won't risk to increase the
836	  memory footprint of applications without a guaranteed
837	  benefit.
838
839	config TRANSPARENT_HUGEPAGE_NEVER
840		bool "never"
841	help
842	  Disable Transparent Hugepage by default. It can still be
843	  enabled at runtime via sysfs.
844endchoice
845
846config THP_SWAP
847	def_bool y
848	depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP && 64BIT
849	help
850	  Swap transparent huge pages in one piece, without splitting.
851	  XXX: For now, swap cluster backing transparent huge page
852	  will be split after swapout.
853
854	  For selection by architectures with reasonable THP sizes.
855
856config READ_ONLY_THP_FOR_FS
857	bool "Read-only THP for filesystems (EXPERIMENTAL)"
858	depends on TRANSPARENT_HUGEPAGE && SHMEM
859
860	help
861	  Allow khugepaged to put read-only file-backed pages in THP.
862
863	  This is marked experimental because it is a new feature. Write
864	  support of file THPs will be developed in the next few release
865	  cycles.
866
867endif # TRANSPARENT_HUGEPAGE
868
869#
870# UP and nommu archs use km based percpu allocator
871#
872config NEED_PER_CPU_KM
873	depends on !SMP || !MMU
874	bool
875	default y
876
877config NEED_PER_CPU_EMBED_FIRST_CHUNK
878	bool
879
880config NEED_PER_CPU_PAGE_FIRST_CHUNK
881	bool
882
883config USE_PERCPU_NUMA_NODE_ID
884	bool
885
886config HAVE_SETUP_PER_CPU_AREA
887	bool
888
889config CMA
890	bool "Contiguous Memory Allocator"
891	depends on MMU
892	select MIGRATION
893	select MEMORY_ISOLATION
894	help
895	  This enables the Contiguous Memory Allocator which allows other
896	  subsystems to allocate big physically-contiguous blocks of memory.
897	  CMA reserves a region of memory and allows only movable pages to
898	  be allocated from it. This way, the kernel can use the memory for
899	  pagecache and when a subsystem requests for contiguous area, the
900	  allocated pages are migrated away to serve the contiguous request.
901
902	  If unsure, say "n".
903
904config CMA_DEBUG
905	bool "CMA debug messages (DEVELOPMENT)"
906	depends on DEBUG_KERNEL && CMA
907	help
908	  Turns on debug messages in CMA.  This produces KERN_DEBUG
909	  messages for every CMA call as well as various messages while
910	  processing calls such as dma_alloc_from_contiguous().
911	  This option does not affect warning and error messages.
912
913config CMA_DEBUGFS
914	bool "CMA debugfs interface"
915	depends on CMA && DEBUG_FS
916	help
917	  Turns on the DebugFS interface for CMA.
918
919config CMA_SYSFS
920	bool "CMA information through sysfs interface"
921	depends on CMA && SYSFS
922	help
923	  This option exposes some sysfs attributes to get information
924	  from CMA.
925
926config CMA_AREAS
927	int "Maximum count of the CMA areas"
928	depends on CMA
929	default 19 if NUMA
930	default 7
931	help
932	  CMA allows to create CMA areas for particular purpose, mainly,
933	  used as device private area. This parameter sets the maximum
934	  number of CMA area in the system.
935
936	  If unsure, leave the default value "7" in UMA and "19" in NUMA.
937
938config MEM_SOFT_DIRTY
939	bool "Track memory changes"
940	depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS
941	select PROC_PAGE_MONITOR
942	help
943	  This option enables memory changes tracking by introducing a
944	  soft-dirty bit on pte-s. This bit it set when someone writes
945	  into a page just as regular dirty bit, but unlike the latter
946	  it can be cleared by hands.
947
948	  See Documentation/admin-guide/mm/soft-dirty.rst for more details.
949
950config GENERIC_EARLY_IOREMAP
951	bool
952
953config STACK_MAX_DEFAULT_SIZE_MB
954	int "Default maximum user stack size for 32-bit processes (MB)"
955	default 100
956	range 8 2048
957	depends on STACK_GROWSUP && (!64BIT || COMPAT)
958	help
959	  This is the maximum stack size in Megabytes in the VM layout of 32-bit
960	  user processes when the stack grows upwards (currently only on parisc
961	  arch) when the RLIMIT_STACK hard limit is unlimited.
962
963	  A sane initial value is 100 MB.
964
965config DEFERRED_STRUCT_PAGE_INIT
966	bool "Defer initialisation of struct pages to kthreads"
967	depends on SPARSEMEM
968	depends on !NEED_PER_CPU_KM
969	depends on 64BIT
970	select PADATA
971	help
972	  Ordinarily all struct pages are initialised during early boot in a
973	  single thread. On very large machines this can take a considerable
974	  amount of time. If this option is set, large machines will bring up
975	  a subset of memmap at boot and then initialise the rest in parallel.
976	  This has a potential performance impact on tasks running early in the
977	  lifetime of the system until these kthreads finish the
978	  initialisation.
979
980config PAGE_IDLE_FLAG
981	bool
982	select PAGE_EXTENSION if !64BIT
983	help
984	  This adds PG_idle and PG_young flags to 'struct page'.  PTE Accessed
985	  bit writers can set the state of the bit in the flags so that PTE
986	  Accessed bit readers may avoid disturbance.
987
988config IDLE_PAGE_TRACKING
989	bool "Enable idle page tracking"
990	depends on SYSFS && MMU
991	select PAGE_IDLE_FLAG
992	help
993	  This feature allows to estimate the amount of user pages that have
994	  not been touched during a given period of time. This information can
995	  be useful to tune memory cgroup limits and/or for job placement
996	  within a compute cluster.
997
998	  See Documentation/admin-guide/mm/idle_page_tracking.rst for
999	  more details.
1000
1001config ARCH_HAS_CACHE_LINE_SIZE
1002	bool
1003
1004config ARCH_HAS_CURRENT_STACK_POINTER
1005	bool
1006	help
1007	  In support of HARDENED_USERCOPY performing stack variable lifetime
1008	  checking, an architecture-agnostic way to find the stack pointer
1009	  is needed. Once an architecture defines an unsigned long global
1010	  register alias named "current_stack_pointer", this config can be
1011	  selected.
1012
1013config ARCH_HAS_PTE_DEVMAP
1014	bool
1015
1016config ARCH_HAS_ZONE_DMA_SET
1017	bool
1018
1019config ZONE_DMA
1020	bool "Support DMA zone" if ARCH_HAS_ZONE_DMA_SET
1021	default y if ARM64 || X86
1022
1023config ZONE_DMA32
1024	bool "Support DMA32 zone" if ARCH_HAS_ZONE_DMA_SET
1025	depends on !X86_32
1026	default y if ARM64
1027
1028config ZONE_DEVICE
1029	bool "Device memory (pmem, HMM, etc...) hotplug support"
1030	depends on MEMORY_HOTPLUG
1031	depends on MEMORY_HOTREMOVE
1032	depends on SPARSEMEM_VMEMMAP
1033	depends on ARCH_HAS_PTE_DEVMAP
1034	select XARRAY_MULTI
1035
1036	help
1037	  Device memory hotplug support allows for establishing pmem,
1038	  or other device driver discovered memory regions, in the
1039	  memmap. This allows pfn_to_page() lookups of otherwise
1040	  "device-physical" addresses which is needed for using a DAX
1041	  mapping in an O_DIRECT operation, among other things.
1042
1043	  If FS_DAX is enabled, then say Y.
1044
1045#
1046# Helpers to mirror range of the CPU page tables of a process into device page
1047# tables.
1048#
1049config HMM_MIRROR
1050	bool
1051	depends on MMU
1052
1053config GET_FREE_REGION
1054	depends on SPARSEMEM
1055	bool
1056
1057config DEVICE_PRIVATE
1058	bool "Unaddressable device memory (GPU memory, ...)"
1059	depends on ZONE_DEVICE
1060	select GET_FREE_REGION
1061
1062	help
1063	  Allows creation of struct pages to represent unaddressable device
1064	  memory; i.e., memory that is only accessible from the device (or
1065	  group of devices). You likely also want to select HMM_MIRROR.
1066
1067config VMAP_PFN
1068	bool
1069
1070config ARCH_USES_HIGH_VMA_FLAGS
1071	bool
1072config ARCH_HAS_PKEYS
1073	bool
1074
1075config ARCH_USES_PG_ARCH_X
1076	bool
1077	help
1078	  Enable the definition of PG_arch_x page flags with x > 1. Only
1079	  suitable for 64-bit architectures with CONFIG_FLATMEM or
1080	  CONFIG_SPARSEMEM_VMEMMAP enabled, otherwise there may not be
1081	  enough room for additional bits in page->flags.
1082
1083config VM_EVENT_COUNTERS
1084	default y
1085	bool "Enable VM event counters for /proc/vmstat" if EXPERT
1086	help
1087	  VM event counters are needed for event counts to be shown.
1088	  This option allows the disabling of the VM event counters
1089	  on EXPERT systems.  /proc/vmstat will only show page counts
1090	  if VM event counters are disabled.
1091
1092config PERCPU_STATS
1093	bool "Collect percpu memory statistics"
1094	help
1095	  This feature collects and exposes statistics via debugfs. The
1096	  information includes global and per chunk statistics, which can
1097	  be used to help understand percpu memory usage.
1098
1099config GUP_TEST
1100	bool "Enable infrastructure for get_user_pages()-related unit tests"
1101	depends on DEBUG_FS
1102	help
1103	  Provides /sys/kernel/debug/gup_test, which in turn provides a way
1104	  to make ioctl calls that can launch kernel-based unit tests for
1105	  the get_user_pages*() and pin_user_pages*() family of API calls.
1106
1107	  These tests include benchmark testing of the _fast variants of
1108	  get_user_pages*() and pin_user_pages*(), as well as smoke tests of
1109	  the non-_fast variants.
1110
1111	  There is also a sub-test that allows running dump_page() on any
1112	  of up to eight pages (selected by command line args) within the
1113	  range of user-space addresses. These pages are either pinned via
1114	  pin_user_pages*(), or pinned via get_user_pages*(), as specified
1115	  by other command line arguments.
1116
1117	  See tools/testing/selftests/mm/gup_test.c
1118
1119comment "GUP_TEST needs to have DEBUG_FS enabled"
1120	depends on !GUP_TEST && !DEBUG_FS
1121
1122config GUP_GET_PXX_LOW_HIGH
1123	bool
1124
1125config DMAPOOL_TEST
1126	tristate "Enable a module to run time tests on dma_pool"
1127	depends on HAS_DMA
1128	help
1129	  Provides a test module that will allocate and free many blocks of
1130	  various sizes and report how long it takes. This is intended to
1131	  provide a consistent way to measure how changes to the
1132	  dma_pool_alloc/free routines affect performance.
1133
1134config ARCH_HAS_PTE_SPECIAL
1135	bool
1136
1137#
1138# Some architectures require a special hugepage directory format that is
1139# required to support multiple hugepage sizes. For example a4fe3ce76
1140# "powerpc/mm: Allow more flexible layouts for hugepage pagetables"
1141# introduced it on powerpc.  This allows for a more flexible hugepage
1142# pagetable layouts.
1143#
1144config ARCH_HAS_HUGEPD
1145	bool
1146
1147config MAPPING_DIRTY_HELPERS
1148        bool
1149
1150config KMAP_LOCAL
1151	bool
1152
1153config KMAP_LOCAL_NON_LINEAR_PTE_ARRAY
1154	bool
1155
1156# struct io_mapping based helper.  Selected by drivers that need them
1157config IO_MAPPING
1158	bool
1159
1160config MEMFD_CREATE
1161	bool "Enable memfd_create() system call" if EXPERT
1162
1163config SECRETMEM
1164	default y
1165	bool "Enable memfd_secret() system call" if EXPERT
1166	depends on ARCH_HAS_SET_DIRECT_MAP
1167	help
1168	  Enable the memfd_secret() system call with the ability to create
1169	  memory areas visible only in the context of the owning process and
1170	  not mapped to other processes and other kernel page tables.
1171
1172config ANON_VMA_NAME
1173	bool "Anonymous VMA name support"
1174	depends on PROC_FS && ADVISE_SYSCALLS && MMU
1175
1176	help
1177	  Allow naming anonymous virtual memory areas.
1178
1179	  This feature allows assigning names to virtual memory areas. Assigned
1180	  names can be later retrieved from /proc/pid/maps and /proc/pid/smaps
1181	  and help identifying individual anonymous memory areas.
1182	  Assigning a name to anonymous virtual memory area might prevent that
1183	  area from being merged with adjacent virtual memory areas due to the
1184	  difference in their name.
1185
1186config HAVE_ARCH_USERFAULTFD_WP
1187	bool
1188	help
1189	  Arch has userfaultfd write protection support
1190
1191config HAVE_ARCH_USERFAULTFD_MINOR
1192	bool
1193	help
1194	  Arch has userfaultfd minor fault support
1195
1196menuconfig USERFAULTFD
1197	bool "Enable userfaultfd() system call"
1198	depends on MMU
1199	help
1200	  Enable the userfaultfd() system call that allows to intercept and
1201	  handle page faults in userland.
1202
1203if USERFAULTFD
1204config PTE_MARKER_UFFD_WP
1205	bool "Userfaultfd write protection support for shmem/hugetlbfs"
1206	default y
1207	depends on HAVE_ARCH_USERFAULTFD_WP
1208
1209	help
1210	  Allows to create marker PTEs for userfaultfd write protection
1211	  purposes.  It is required to enable userfaultfd write protection on
1212	  file-backed memory types like shmem and hugetlbfs.
1213endif # USERFAULTFD
1214
1215# multi-gen LRU {
1216config LRU_GEN
1217	bool "Multi-Gen LRU"
1218	depends on MMU
1219	# make sure folio->flags has enough spare bits
1220	depends on 64BIT || !SPARSEMEM || SPARSEMEM_VMEMMAP
1221	help
1222	  A high performance LRU implementation to overcommit memory. See
1223	  Documentation/admin-guide/mm/multigen_lru.rst for details.
1224
1225config LRU_GEN_ENABLED
1226	bool "Enable by default"
1227	depends on LRU_GEN
1228	help
1229	  This option enables the multi-gen LRU by default.
1230
1231config LRU_GEN_STATS
1232	bool "Full stats for debugging"
1233	depends on LRU_GEN
1234	help
1235	  Do not enable this option unless you plan to look at historical stats
1236	  from evicted generations for debugging purpose.
1237
1238	  This option has a per-memcg and per-node memory overhead.
1239
1240config LRU_GEN_WALKS_MMU
1241	def_bool y
1242	depends on LRU_GEN && ARCH_HAS_HW_PTE_YOUNG
1243# }
1244
1245config ARCH_SUPPORTS_PER_VMA_LOCK
1246       def_bool n
1247
1248config PER_VMA_LOCK
1249	def_bool y
1250	depends on ARCH_SUPPORTS_PER_VMA_LOCK && MMU && SMP
1251	help
1252	  Allow per-vma locking during page fault handling.
1253
1254	  This feature allows locking each virtual memory area separately when
1255	  handling page faults instead of taking mmap_lock.
1256
1257config LOCK_MM_AND_FIND_VMA
1258	bool
1259	depends on !STACK_GROWSUP
1260
1261source "mm/damon/Kconfig"
1262
1263endmenu
1264