1 # SPDX-License-Identifier: GPL-2.0-only
3 menu "Memory Management options"
5 config SELECT_MEMORY_MODEL
7 depends on ARCH_SELECT_MEMORY_MODEL
11 depends on SELECT_MEMORY_MODEL
12 default DISCONTIGMEM_MANUAL if ARCH_DISCONTIGMEM_DEFAULT
13 default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT
14 default FLATMEM_MANUAL
16 This option allows you to change some of the ways that
17 Linux manages its memory internally. Most users will
18 only have one option here selected by the architecture
19 configuration. This is normal.
23 depends on !(ARCH_DISCONTIGMEM_ENABLE || ARCH_SPARSEMEM_ENABLE) || ARCH_FLATMEM_ENABLE
25 This option is best suited for non-NUMA systems with
26 flat address space. The FLATMEM is the most efficient
27 system in terms of performance and resource consumption
28 and it is the best option for smaller systems.
30 For systems that have holes in their physical address
31 spaces and for features like NUMA and memory hotplug,
32 choose "Sparse Memory"
34 If unsure, choose this option (Flat Memory) over any other.
36 config DISCONTIGMEM_MANUAL
37 bool "Discontiguous Memory"
38 depends on ARCH_DISCONTIGMEM_ENABLE
40 This option provides enhanced support for discontiguous
41 memory systems, over FLATMEM. These systems have holes
42 in their physical address spaces, and this option provides
43 more efficient handling of these holes.
45 Although "Discontiguous Memory" is still used by several
46 architectures, it is considered deprecated in favor of
49 If unsure, choose "Sparse Memory" over this option.
51 config SPARSEMEM_MANUAL
53 depends on ARCH_SPARSEMEM_ENABLE
55 This will be the only option for some systems, including
56 memory hot-plug systems. This is normal.
58 This option provides efficient support for systems with
59 holes is their physical address space and allows memory
60 hot-plug and hot-remove.
62 If unsure, choose "Flat Memory" over this option.
68 depends on (!SELECT_MEMORY_MODEL && ARCH_DISCONTIGMEM_ENABLE) || DISCONTIGMEM_MANUAL
72 depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL
76 depends on (!DISCONTIGMEM && !SPARSEMEM) || FLATMEM_MANUAL
78 config FLAT_NODE_MEM_MAP
83 # Both the NUMA code and DISCONTIGMEM use arrays of pg_data_t's
84 # to represent different areas of memory. This variable allows
85 # those dependencies to exist individually.
87 config NEED_MULTIPLE_NODES
89 depends on DISCONTIGMEM || NUMA
91 config HAVE_MEMORY_PRESENT
93 depends on ARCH_HAVE_MEMORY_PRESENT || SPARSEMEM
96 # SPARSEMEM_EXTREME (which is the default) does some bootmem
97 # allocations when memory_present() is called. If this cannot
98 # be done on your architecture, select this option. However,
99 # statically allocating the mem_section[] array can potentially
100 # consume vast quantities of .bss, so be careful.
102 # This option will also potentially produce smaller runtime code
103 # with gcc 3.4 and later.
105 config SPARSEMEM_STATIC
109 # Architecture platforms which require a two level mem_section in SPARSEMEM
110 # must select this option. This is usually for architecture platforms with
111 # an extremely sparse physical address space.
113 config SPARSEMEM_EXTREME
115 depends on SPARSEMEM && !SPARSEMEM_STATIC
117 config SPARSEMEM_VMEMMAP_ENABLE
120 config SPARSEMEM_VMEMMAP
121 bool "Sparse Memory virtual memmap"
122 depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE
125 SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise
126 pfn_to_page and page_to_pfn operations. This is the most
127 efficient option when sufficient kernel resources are available.
129 config HAVE_MEMBLOCK_NODE_MAP
132 config HAVE_MEMBLOCK_PHYS_MAP
139 config ARCH_KEEP_MEMBLOCK
142 config MEMORY_ISOLATION
146 # Only be set on architectures that have completely implemented memory hotplug
147 # feature. If you are not sure, don't touch it.
149 config HAVE_BOOTMEM_INFO_NODE
152 # eventually, we can have this option just 'select SPARSEMEM'
153 config MEMORY_HOTPLUG
154 bool "Allow for memory hot-add"
155 depends on SPARSEMEM || X86_64_ACPI_NUMA
156 depends on ARCH_ENABLE_MEMORY_HOTPLUG
158 config MEMORY_HOTPLUG_SPARSE
160 depends on SPARSEMEM && MEMORY_HOTPLUG
162 config MEMORY_HOTPLUG_DEFAULT_ONLINE
163 bool "Online the newly added memory blocks by default"
164 depends on MEMORY_HOTPLUG
166 This option sets the default policy setting for memory hotplug
167 onlining policy (/sys/devices/system/memory/auto_online_blocks) which
168 determines what happens to newly added memory regions. Policy setting
169 can always be changed at runtime.
170 See Documentation/admin-guide/mm/memory-hotplug.rst for more information.
172 Say Y here if you want all hot-plugged memory blocks to appear in
173 'online' state by default.
174 Say N here if you want the default policy to keep all hot-plugged
175 memory blocks in 'offline' state.
177 config MEMORY_HOTREMOVE
178 bool "Allow for memory hot remove"
179 select MEMORY_ISOLATION
180 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64)
181 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
184 # Heavily threaded applications may benefit from splitting the mm-wide
185 # page_table_lock, so that faults on different parts of the user address
186 # space can be handled with less contention: split it at this NR_CPUS.
187 # Default to 4 for wider testing, though 8 might be more appropriate.
188 # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
189 # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
190 # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page.
192 config SPLIT_PTLOCK_CPUS
194 default "999999" if !MMU
195 default "999999" if ARM && !CPU_CACHE_VIPT
196 default "999999" if PARISC && !PA20
199 config ARCH_ENABLE_SPLIT_PMD_PTLOCK
203 # support for memory balloon
204 config MEMORY_BALLOON
208 # support for memory balloon compaction
209 config BALLOON_COMPACTION
210 bool "Allow for balloon memory compaction/migration"
212 depends on COMPACTION && MEMORY_BALLOON
214 Memory fragmentation introduced by ballooning might reduce
215 significantly the number of 2MB contiguous memory blocks that can be
216 used within a guest, thus imposing performance penalties associated
217 with the reduced number of transparent huge pages that could be used
218 by the guest workload. Allowing the compaction & migration for memory
219 pages enlisted as being part of memory balloon devices avoids the
220 scenario aforementioned and helps improving memory defragmentation.
223 # support for memory compaction
225 bool "Allow for memory compaction"
230 Compaction is the only memory management component to form
231 high order (larger physically contiguous) memory blocks
232 reliably. The page allocator relies on compaction heavily and
233 the lack of the feature can lead to unexpected OOM killer
234 invocations for high order memory requests. You shouldn't
235 disable this option unless there really is a strong reason for
236 it and then we would be really interested to hear about that at
240 # support for page migration
243 bool "Page migration"
245 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU
247 Allows the migration of the physical location of pages of processes
248 while the virtual addresses are not changed. This is useful in
249 two situations. The first is on NUMA systems to put pages nearer
250 to the processors accessing. The second is when allocating huge
251 pages as migration can relocate pages to satisfy a huge page
252 allocation instead of reclaiming.
254 config ARCH_ENABLE_HUGEPAGE_MIGRATION
257 config ARCH_ENABLE_THP_MIGRATION
261 def_bool (MEMORY_ISOLATION && COMPACTION) || CMA
263 config PHYS_ADDR_T_64BIT
267 bool "Enable bounce buffers"
269 depends on BLOCK && MMU && (ZONE_DMA || HIGHMEM)
271 Enable bounce buffers for devices that cannot access
272 the full range of memory available to the CPU. Enabled
273 by default when ZONE_DMA or HIGHMEM is selected, but you
274 may say n to override this.
279 An architecture should select this if it implements the
280 deprecated interface virt_to_bus(). All new architectures
281 should probably not select this.
289 bool "Enable KSM for page merging"
293 Enable Kernel Samepage Merging: KSM periodically scans those areas
294 of an application's address space that an app has advised may be
295 mergeable. When it finds pages of identical content, it replaces
296 the many instances by a single page with that content, so
297 saving memory until one or another app needs to modify the content.
298 Recommended for use with KVM, or with other duplicative applications.
299 See Documentation/vm/ksm.rst for more information: KSM is inactive
300 until a program has madvised that an area is MADV_MERGEABLE, and
301 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
304 bool "Enable lightweight KSM"
308 LKSM is intended to mitigate energy cost of KSM. LKSM sleeps and
309 waits for incoming events, such as freeze of tasks and user-level
312 Then, LKSM performs one of two memory deduplication strategies, such as
313 partial deduplication and full deduplication.
314 As names of them implying, partial deduplication only scans a subset of
315 processes, thus it provides a lightweight way. Full deduplication scans
316 whole processes joined in the KSM scanning list, thus it is same to
317 original KSM. Due to the level of the incurred event, LKSM performs
318 a proper deduplication strategy.
321 bool "Enable LKSM filtering feature"
324 LKSM_FILTER is intended to reduce necessary scans. LKSM records merged
325 addresses and then used it to filter out unmergeable pages.
326 It improves energy-efficiency at the expense of somewhat sharing chances.
328 config DEFAULT_MMAP_MIN_ADDR
329 int "Low address space to protect from user allocation"
333 This is the portion of low virtual memory which should be protected
334 from userspace allocation. Keeping a user from writing to low pages
335 can help reduce the impact of kernel NULL pointer bugs.
337 For most ia64, ppc64 and x86 users with lots of address space
338 a value of 65536 is reasonable and should cause no problems.
339 On arm and other archs it should not be higher than 32768.
340 Programs which use vm86 functionality or have some need to map
341 this low address space will need CAP_SYS_RAWIO or disable this
342 protection by setting the value to 0.
344 This value can be changed after boot using the
345 /proc/sys/vm/mmap_min_addr tunable.
347 config ARCH_SUPPORTS_MEMORY_FAILURE
350 config MEMORY_FAILURE
352 depends on ARCH_SUPPORTS_MEMORY_FAILURE
353 bool "Enable recovery from hardware memory errors"
354 select MEMORY_ISOLATION
357 Enables code to recover from some memory failures on systems
358 with MCA recovery. This allows a system to continue running
359 even when some of its memory has uncorrected errors. This requires
360 special hardware support and typically ECC memory.
362 config HWPOISON_INJECT
363 tristate "HWPoison pages injector"
364 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS
365 select PROC_PAGE_MONITOR
367 config NOMMU_INITIAL_TRIM_EXCESS
368 int "Turn on mmap() excess space trimming before booting"
372 The NOMMU mmap() frequently needs to allocate large contiguous chunks
373 of memory on which to store mappings, but it can only ask the system
374 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
375 more than it requires. To deal with this, mmap() is able to trim off
376 the excess and return it to the allocator.
378 If trimming is enabled, the excess is trimmed off and returned to the
379 system allocator, which can cause extra fragmentation, particularly
380 if there are a lot of transient processes.
382 If trimming is disabled, the excess is kept, but not used, which for
383 long-term mappings means that the space is wasted.
385 Trimming can be dynamically controlled through a sysctl option
386 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of
387 excess pages there must be before trimming should occur, or zero if
388 no trimming is to occur.
390 This option specifies the initial value of this option. The default
391 of 1 says that all excess pages should be trimmed.
393 See Documentation/nommu-mmap.txt for more information.
395 config TRANSPARENT_HUGEPAGE
396 bool "Transparent Hugepage Support"
397 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE
401 Transparent Hugepages allows the kernel to use huge pages and
402 huge tlb transparently to the applications whenever possible.
403 This feature can improve computing performance to certain
404 applications by speeding up page faults during memory
405 allocation, by reducing the number of tlb misses and by speeding
406 up the pagetable walking.
408 If memory constrained on embedded, you may want to say N.
411 prompt "Transparent Hugepage Support sysfs defaults"
412 depends on TRANSPARENT_HUGEPAGE
413 default TRANSPARENT_HUGEPAGE_ALWAYS
415 Selects the sysfs defaults for Transparent Hugepage Support.
417 config TRANSPARENT_HUGEPAGE_ALWAYS
420 Enabling Transparent Hugepage always, can increase the
421 memory footprint of applications without a guaranteed
422 benefit but it will work automatically for all applications.
424 config TRANSPARENT_HUGEPAGE_MADVISE
427 Enabling Transparent Hugepage madvise, will only provide a
428 performance improvement benefit to the applications using
429 madvise(MADV_HUGEPAGE) but it won't risk to increase the
430 memory footprint of applications without a guaranteed
434 config ARCH_WANTS_THP_SWAP
439 depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP
441 Swap transparent huge pages in one piece, without splitting.
442 XXX: For now, swap cluster backing transparent huge page
443 will be split after swapout.
445 For selection by architectures with reasonable THP sizes.
447 config TRANSPARENT_HUGE_PAGECACHE
449 depends on TRANSPARENT_HUGEPAGE
452 # UP and nommu archs use km based percpu allocator
454 config NEED_PER_CPU_KM
460 bool "Enable cleancache driver to cache clean pages if tmem is present"
462 Cleancache can be thought of as a page-granularity victim cache
463 for clean pages that the kernel's pageframe replacement algorithm
464 (PFRA) would like to keep around, but can't since there isn't enough
465 memory. So when the PFRA "evicts" a page, it first attempts to use
466 cleancache code to put the data contained in that page into
467 "transcendent memory", memory that is not directly accessible or
468 addressable by the kernel and is of unknown and possibly
469 time-varying size. And when a cleancache-enabled
470 filesystem wishes to access a page in a file on disk, it first
471 checks cleancache to see if it already contains it; if it does,
472 the page is copied into the kernel and a disk access is avoided.
473 When a transcendent memory driver is available (such as zcache or
474 Xen transcendent memory), a significant I/O reduction
475 may be achieved. When none is available, all cleancache calls
476 are reduced to a single pointer-compare-against-NULL resulting
477 in a negligible performance hit.
479 If unsure, say Y to enable cleancache
482 bool "Enable frontswap to cache swap pages if tmem is present"
485 Frontswap is so named because it can be thought of as the opposite
486 of a "backing" store for a swap device. The data is stored into
487 "transcendent memory", memory that is not directly accessible or
488 addressable by the kernel and is of unknown and possibly
489 time-varying size. When space in transcendent memory is available,
490 a significant swap I/O reduction may be achieved. When none is
491 available, all frontswap calls are reduced to a single pointer-
492 compare-against-NULL resulting in a negligible performance hit
493 and swap data is stored as normal on the matching swap device.
495 If unsure, say Y to enable frontswap.
498 bool "Contiguous Memory Allocator"
501 select MEMORY_ISOLATION
503 This enables the Contiguous Memory Allocator which allows other
504 subsystems to allocate big physically-contiguous blocks of memory.
505 CMA reserves a region of memory and allows only movable pages to
506 be allocated from it. This way, the kernel can use the memory for
507 pagecache and when a subsystem requests for contiguous area, the
508 allocated pages are migrated away to serve the contiguous request.
513 bool "CMA debug messages (DEVELOPMENT)"
514 depends on DEBUG_KERNEL && CMA
516 Turns on debug messages in CMA. This produces KERN_DEBUG
517 messages for every CMA call as well as various messages while
518 processing calls such as dma_alloc_from_contiguous().
519 This option does not affect warning and error messages.
522 bool "CMA debugfs interface"
523 depends on CMA && DEBUG_FS
525 Turns on the DebugFS interface for CMA.
528 int "Maximum count of the CMA areas"
532 CMA allows to create CMA areas for particular purpose, mainly,
533 used as device private area. This parameter sets the maximum
534 number of CMA area in the system.
536 If unsure, leave the default value "7".
538 config MEM_SOFT_DIRTY
539 bool "Track memory changes"
540 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS
541 select PROC_PAGE_MONITOR
543 This option enables memory changes tracking by introducing a
544 soft-dirty bit on pte-s. This bit it set when someone writes
545 into a page just as regular dirty bit, but unlike the latter
546 it can be cleared by hands.
548 See Documentation/admin-guide/mm/soft-dirty.rst for more details.
551 bool "Compressed cache for swap pages (EXPERIMENTAL)"
552 depends on FRONTSWAP && CRYPTO=y
556 A lightweight compressed cache for swap pages. It takes
557 pages that are in the process of being swapped out and attempts to
558 compress them into a dynamically allocated RAM-based memory pool.
559 This can result in a significant I/O reduction on swap device and,
560 in the case where decompressing from RAM is faster that swap device
561 reads, can also improve workload performance.
563 This is marked experimental because it is a new feature (as of
564 v3.11) that interacts heavily with memory reclaim. While these
565 interactions don't cause any known issues on simple memory setups,
566 they have not be fully explored on the large set of potential
567 configurations and workloads that exist.
570 tristate "Common API for compressed memory storage"
572 Compressed memory storage API. This allows using either zbud or
576 tristate "Low (Up to 2x) density storage for compressed pages"
578 A special purpose allocator for storing compressed pages.
579 It is designed to store up to two compressed pages per physical
580 page. While this design limits storage density, it has simple and
581 deterministic reclaim properties that make it preferable to a higher
582 density approach when reclaim will be used.
585 tristate "Up to 3x density storage for compressed pages"
588 A special purpose allocator for storing compressed pages.
589 It is designed to store up to three compressed pages per physical
590 page. It is a ZBUD derivative so the simplicity and determinism are
594 tristate "Memory allocator for compressed pages"
597 zsmalloc is a slab-based memory allocator designed to store
598 compressed RAM pages. zsmalloc uses virtual memory mapping
599 in order to reduce fragmentation. However, this results in a
600 non-standard allocator interface where a handle, not a pointer, is
601 returned by an alloc(). This handle must be mapped in order to
602 access the allocated space.
604 config PGTABLE_MAPPING
605 bool "Use page table mapping to access object in zsmalloc"
608 By default, zsmalloc uses a copy-based object mapping method to
609 access allocations that span two pages. However, if a particular
610 architecture (ex, ARM) performs VM mapping faster than copying,
611 then you should select this. This causes zsmalloc to use page table
612 mapping rather than copying for object mapping.
614 You can check speed with zsmalloc benchmark:
615 https://github.com/spartacus06/zsmapbench
618 bool "Export zsmalloc statistics"
622 This option enables code in the zsmalloc to collect various
623 statistics about whats happening in zsmalloc and exports that
624 information to userspace via debugfs.
627 config GENERIC_EARLY_IOREMAP
630 config MAX_STACK_SIZE_MB
631 int "Maximum user stack size for 32-bit processes (MB)"
634 depends on STACK_GROWSUP && (!64BIT || COMPAT)
636 This is the maximum stack size in Megabytes in the VM layout of 32-bit
637 user processes when the stack grows upwards (currently only on parisc
638 arch). The stack will be located at the highest memory address minus
639 the given value, unless the RLIMIT_STACK hard limit is changed to a
640 smaller value in which case that is used.
642 A sane initial value is 80 MB.
644 config DEFERRED_STRUCT_PAGE_INIT
645 bool "Defer initialisation of struct pages to kthreads"
647 depends on !NEED_PER_CPU_KM
650 Ordinarily all struct pages are initialised during early boot in a
651 single thread. On very large machines this can take a considerable
652 amount of time. If this option is set, large machines will bring up
653 a subset of memmap at boot and then initialise the rest in parallel
654 by starting one-off "pgdatinitX" kernel thread for each node X. This
655 has a potential performance impact on processes running early in the
656 lifetime of the system until these kthreads finish the
659 config IDLE_PAGE_TRACKING
660 bool "Enable idle page tracking"
661 depends on SYSFS && MMU
662 select PAGE_EXTENSION if !64BIT
664 This feature allows to estimate the amount of user pages that have
665 not been touched during a given period of time. This information can
666 be useful to tune memory cgroup limits and/or for job placement
667 within a compute cluster.
669 See Documentation/admin-guide/mm/idle_page_tracking.rst for
672 config ARCH_HAS_PTE_DEVMAP
676 bool "Device memory (pmem, HMM, etc...) hotplug support"
677 depends on MEMORY_HOTPLUG
678 depends on MEMORY_HOTREMOVE
679 depends on SPARSEMEM_VMEMMAP
680 depends on ARCH_HAS_PTE_DEVMAP
684 Device memory hotplug support allows for establishing pmem,
685 or other device driver discovered memory regions, in the
686 memmap. This allows pfn_to_page() lookups of otherwise
687 "device-physical" addresses which is needed for using a DAX
688 mapping in an O_DIRECT operation, among other things.
690 If FS_DAX is enabled, then say Y.
692 config DEV_PAGEMAP_OPS
696 # Helpers to mirror range of the CPU page tables of a process into device page
702 depends on MMU_NOTIFIER
704 config DEVICE_PRIVATE
705 bool "Unaddressable device memory (GPU memory, ...)"
706 depends on ZONE_DEVICE
707 select DEV_PAGEMAP_OPS
710 Allows creation of struct pages to represent unaddressable device
711 memory; i.e., memory that is only accessible from the device (or
712 group of devices). You likely also want to select HMM_MIRROR.
717 config ARCH_USES_HIGH_VMA_FLAGS
719 config ARCH_HAS_PKEYS
723 bool "Collect percpu memory statistics"
725 This feature collects and exposes statistics via debugfs. The
726 information includes global and per chunk statistics, which can
727 be used to help understand percpu memory usage.
730 bool "Enable infrastructure for get_user_pages_fast() benchmarking"
732 Provides /sys/kernel/debug/gup_benchmark that helps with testing
733 performance of get_user_pages_fast().
735 See tools/testing/selftests/vm/gup_benchmark.c
737 config GUP_GET_PTE_LOW_HIGH
740 config READ_ONLY_THP_FOR_FS
741 bool "Read-only THP for filesystems (EXPERIMENTAL)"
742 depends on TRANSPARENT_HUGE_PAGECACHE && SHMEM
745 Allow khugepaged to put read-only file-backed pages in THP.
747 This is marked experimental because it is a new feature. Write
748 support of file THPs will be developed in the next few release
751 config ARCH_HAS_PTE_SPECIAL
755 # Some architectures require a special hugepage directory format that is
756 # required to support multiple hugepage sizes. For example a4fe3ce76
757 # "powerpc/mm: Allow more flexible layouts for hugepage pagetables"
758 # introduced it on powerpc. This allows for a more flexible hugepage
761 config ARCH_HAS_HUGEPD