1 # SPDX-License-Identifier: GPL-2.0-only
3 menu "Memory Management options"
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.
16 bool "Support for paging of anonymous memory (swap)"
17 depends on MMU && BLOCK && !ARCH_NO_SWAP
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.
26 bool "Compressed cache for swap pages"
32 A lightweight compressed cache for swap pages. It takes
33 pages that are in the process of being swapped out and attempts to
34 compress them into a dynamically allocated RAM-based memory pool.
35 This can result in a significant I/O reduction on swap device and,
36 in the case where decompressing from RAM is faster than swap device
37 reads, can also improve workload performance.
39 config ZSWAP_DEFAULT_ON
40 bool "Enable the compressed cache for swap pages by default"
43 If selected, the compressed cache for swap pages will be enabled
44 at boot, otherwise it will be disabled.
46 The selection made here can be overridden by using the kernel
47 command line 'zswap.enabled=' option.
50 prompt "Default compressor"
52 default ZSWAP_COMPRESSOR_DEFAULT_LZO
54 Selects the default compression algorithm for the compressed cache
57 For an overview what kind of performance can be expected from
58 a particular compression algorithm please refer to the benchmarks
59 available at the following LWN page:
60 https://lwn.net/Articles/751795/
62 If in doubt, select 'LZO'.
64 The selection made here can be overridden by using the kernel
65 command line 'zswap.compressor=' option.
67 config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
71 Use the Deflate algorithm as the default compression algorithm.
73 config ZSWAP_COMPRESSOR_DEFAULT_LZO
77 Use the LZO algorithm as the default compression algorithm.
79 config ZSWAP_COMPRESSOR_DEFAULT_842
83 Use the 842 algorithm as the default compression algorithm.
85 config ZSWAP_COMPRESSOR_DEFAULT_LZ4
89 Use the LZ4 algorithm as the default compression algorithm.
91 config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
95 Use the LZ4HC algorithm as the default compression algorithm.
97 config ZSWAP_COMPRESSOR_DEFAULT_ZSTD
101 Use the zstd algorithm as the default compression algorithm.
104 config ZSWAP_COMPRESSOR_DEFAULT
107 default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
108 default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO
109 default "842" if ZSWAP_COMPRESSOR_DEFAULT_842
110 default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4
111 default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
112 default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD
116 prompt "Default allocator"
118 default ZSWAP_ZPOOL_DEFAULT_ZBUD
120 Selects the default allocator for the compressed cache for
122 The default is 'zbud' for compatibility, however please do
123 read the description of each of the allocators below before
124 making a right choice.
126 The selection made here can be overridden by using the kernel
127 command line 'zswap.zpool=' option.
129 config ZSWAP_ZPOOL_DEFAULT_ZBUD
133 Use the zbud allocator as the default allocator.
135 config ZSWAP_ZPOOL_DEFAULT_Z3FOLD
139 Use the z3fold allocator as the default allocator.
141 config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
145 Use the zsmalloc allocator as the default allocator.
148 config ZSWAP_ZPOOL_DEFAULT
151 default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD
152 default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD
153 default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
157 tristate "2:1 compression allocator (zbud)"
160 A special purpose allocator for storing compressed pages.
161 It is designed to store up to two compressed pages per physical
162 page. While this design limits storage density, it has simple and
163 deterministic reclaim properties that make it preferable to a higher
164 density approach when reclaim will be used.
167 tristate "3:1 compression allocator (z3fold)"
170 A special purpose allocator for storing compressed pages.
171 It is designed to store up to three compressed pages per physical
172 page. It is a ZBUD derivative so the simplicity and determinism are
177 prompt "N:1 compression allocator (zsmalloc)" if ZSWAP
180 zsmalloc is a slab-based memory allocator designed to store
181 pages of various compression levels efficiently. It achieves
182 the highest storage density with the least amount of fragmentation.
185 bool "Export zsmalloc statistics"
189 This option enables code in the zsmalloc to collect various
190 statistics about what's happening in zsmalloc and exports that
191 information to userspace via debugfs.
194 menu "SLAB allocator options"
197 prompt "Choose SLAB allocator"
200 This option allows to select a slab allocator.
204 depends on !PREEMPT_RT
205 select HAVE_HARDENED_USERCOPY_ALLOCATOR
207 The regular slab allocator that is established and known to work
208 well in all environments. It organizes cache hot objects in
209 per cpu and per node queues.
212 bool "SLUB (Unqueued Allocator)"
213 select HAVE_HARDENED_USERCOPY_ALLOCATOR
215 SLUB is a slab allocator that minimizes cache line usage
216 instead of managing queues of cached objects (SLAB approach).
217 Per cpu caching is realized using slabs of objects instead
218 of queues of objects. SLUB can use memory efficiently
219 and has enhanced diagnostics. SLUB is the default choice for
224 bool "SLOB (Simple Allocator)"
225 depends on !PREEMPT_RT
227 SLOB replaces the stock allocator with a drastically simpler
228 allocator. SLOB is generally more space efficient but
229 does not perform as well on large systems.
233 config SLAB_MERGE_DEFAULT
234 bool "Allow slab caches to be merged"
236 depends on SLAB || SLUB
238 For reduced kernel memory fragmentation, slab caches can be
239 merged when they share the same size and other characteristics.
240 This carries a risk of kernel heap overflows being able to
241 overwrite objects from merged caches (and more easily control
242 cache layout), which makes such heap attacks easier to exploit
243 by attackers. By keeping caches unmerged, these kinds of exploits
244 can usually only damage objects in the same cache. To disable
245 merging at runtime, "slab_nomerge" can be passed on the kernel
248 config SLAB_FREELIST_RANDOM
249 bool "Randomize slab freelist"
250 depends on SLAB || SLUB
252 Randomizes the freelist order used on creating new pages. This
253 security feature reduces the predictability of the kernel slab
254 allocator against heap overflows.
256 config SLAB_FREELIST_HARDENED
257 bool "Harden slab freelist metadata"
258 depends on SLAB || SLUB
260 Many kernel heap attacks try to target slab cache metadata and
261 other infrastructure. This options makes minor performance
262 sacrifices to harden the kernel slab allocator against common
263 freelist exploit methods. Some slab implementations have more
264 sanity-checking than others. This option is most effective with
269 bool "Enable SLUB performance statistics"
270 depends on SLUB && SYSFS
272 SLUB statistics are useful to debug SLUBs allocation behavior in
273 order find ways to optimize the allocator. This should never be
274 enabled for production use since keeping statistics slows down
275 the allocator by a few percentage points. The slabinfo command
276 supports the determination of the most active slabs to figure
277 out which slabs are relevant to a particular load.
278 Try running: slabinfo -DA
280 config SLUB_CPU_PARTIAL
282 depends on SLUB && SMP
283 bool "SLUB per cpu partial cache"
285 Per cpu partial caches accelerate objects allocation and freeing
286 that is local to a processor at the price of more indeterminism
287 in the latency of the free. On overflow these caches will be cleared
288 which requires the taking of locks that may cause latency spikes.
289 Typically one would choose no for a realtime system.
291 endmenu # SLAB allocator options
293 config SHUFFLE_PAGE_ALLOCATOR
294 bool "Page allocator randomization"
295 default SLAB_FREELIST_RANDOM && ACPI_NUMA
297 Randomization of the page allocator improves the average
298 utilization of a direct-mapped memory-side-cache. See section
299 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
300 6.2a specification for an example of how a platform advertises
301 the presence of a memory-side-cache. There are also incidental
302 security benefits as it reduces the predictability of page
303 allocations to compliment SLAB_FREELIST_RANDOM, but the
304 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
305 10th order of pages is selected based on cache utilization
308 While the randomization improves cache utilization it may
309 negatively impact workloads on platforms without a cache. For
310 this reason, by default, the randomization is enabled only
311 after runtime detection of a direct-mapped memory-side-cache.
312 Otherwise, the randomization may be force enabled with the
313 'page_alloc.shuffle' kernel command line parameter.
318 bool "Disable heap randomization"
321 Randomizing heap placement makes heap exploits harder, but it
322 also breaks ancient binaries (including anything libc5 based).
323 This option changes the bootup default to heap randomization
324 disabled, and can be overridden at runtime by setting
325 /proc/sys/kernel/randomize_va_space to 2.
327 On non-ancient distros (post-2000 ones) N is usually a safe choice.
329 config MMAP_ALLOW_UNINITIALIZED
330 bool "Allow mmapped anonymous memory to be uninitialized"
331 depends on EXPERT && !MMU
334 Normally, and according to the Linux spec, anonymous memory obtained
335 from mmap() has its contents cleared before it is passed to
336 userspace. Enabling this config option allows you to request that
337 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
338 providing a huge performance boost. If this option is not enabled,
339 then the flag will be ignored.
341 This is taken advantage of by uClibc's malloc(), and also by
342 ELF-FDPIC binfmt's brk and stack allocator.
344 Because of the obvious security issues, this option should only be
345 enabled on embedded devices where you control what is run in
346 userspace. Since that isn't generally a problem on no-MMU systems,
347 it is normally safe to say Y here.
349 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
351 config SELECT_MEMORY_MODEL
353 depends on ARCH_SELECT_MEMORY_MODEL
356 prompt "Memory model"
357 depends on SELECT_MEMORY_MODEL
358 default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT
359 default FLATMEM_MANUAL
361 This option allows you to change some of the ways that
362 Linux manages its memory internally. Most users will
363 only have one option here selected by the architecture
364 configuration. This is normal.
366 config FLATMEM_MANUAL
368 depends on !ARCH_SPARSEMEM_ENABLE || ARCH_FLATMEM_ENABLE
370 This option is best suited for non-NUMA systems with
371 flat address space. The FLATMEM is the most efficient
372 system in terms of performance and resource consumption
373 and it is the best option for smaller systems.
375 For systems that have holes in their physical address
376 spaces and for features like NUMA and memory hotplug,
377 choose "Sparse Memory".
379 If unsure, choose this option (Flat Memory) over any other.
381 config SPARSEMEM_MANUAL
383 depends on ARCH_SPARSEMEM_ENABLE
385 This will be the only option for some systems, including
386 memory hot-plug systems. This is normal.
388 This option provides efficient support for systems with
389 holes is their physical address space and allows memory
390 hot-plug and hot-remove.
392 If unsure, choose "Flat Memory" over this option.
398 depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL
402 depends on !SPARSEMEM || FLATMEM_MANUAL
405 # SPARSEMEM_EXTREME (which is the default) does some bootmem
406 # allocations when sparse_init() is called. If this cannot
407 # be done on your architecture, select this option. However,
408 # statically allocating the mem_section[] array can potentially
409 # consume vast quantities of .bss, so be careful.
411 # This option will also potentially produce smaller runtime code
412 # with gcc 3.4 and later.
414 config SPARSEMEM_STATIC
418 # Architecture platforms which require a two level mem_section in SPARSEMEM
419 # must select this option. This is usually for architecture platforms with
420 # an extremely sparse physical address space.
422 config SPARSEMEM_EXTREME
424 depends on SPARSEMEM && !SPARSEMEM_STATIC
426 config SPARSEMEM_VMEMMAP_ENABLE
429 config SPARSEMEM_VMEMMAP
430 bool "Sparse Memory virtual memmap"
431 depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE
434 SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise
435 pfn_to_page and page_to_pfn operations. This is the most
436 efficient option when sufficient kernel resources are available.
438 config HAVE_MEMBLOCK_PHYS_MAP
445 # Don't discard allocated memory used to track "memory" and "reserved" memblocks
446 # after early boot, so it can still be used to test for validity of memory.
447 # Also, memblocks are updated with memory hot(un)plug.
448 config ARCH_KEEP_MEMBLOCK
451 # Keep arch NUMA mapping infrastructure post-init.
452 config NUMA_KEEP_MEMINFO
455 config MEMORY_ISOLATION
458 # IORESOURCE_SYSTEM_RAM regions in the kernel resource tree that are marked
459 # IORESOURCE_EXCLUSIVE cannot be mapped to user space, for example, via
461 config EXCLUSIVE_SYSTEM_RAM
463 depends on !DEVMEM || STRICT_DEVMEM
466 # Only be set on architectures that have completely implemented memory hotplug
467 # feature. If you are not sure, don't touch it.
469 config HAVE_BOOTMEM_INFO_NODE
472 config ARCH_ENABLE_MEMORY_HOTPLUG
475 config ARCH_ENABLE_MEMORY_HOTREMOVE
478 # eventually, we can have this option just 'select SPARSEMEM'
479 menuconfig MEMORY_HOTPLUG
480 bool "Memory hotplug"
481 select MEMORY_ISOLATION
483 depends on ARCH_ENABLE_MEMORY_HOTPLUG
485 select NUMA_KEEP_MEMINFO if NUMA
489 config MEMORY_HOTPLUG_DEFAULT_ONLINE
490 bool "Online the newly added memory blocks by default"
491 depends on MEMORY_HOTPLUG
493 This option sets the default policy setting for memory hotplug
494 onlining policy (/sys/devices/system/memory/auto_online_blocks) which
495 determines what happens to newly added memory regions. Policy setting
496 can always be changed at runtime.
497 See Documentation/admin-guide/mm/memory-hotplug.rst for more information.
499 Say Y here if you want all hot-plugged memory blocks to appear in
500 'online' state by default.
501 Say N here if you want the default policy to keep all hot-plugged
502 memory blocks in 'offline' state.
504 config MEMORY_HOTREMOVE
505 bool "Allow for memory hot remove"
506 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64)
507 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
510 config MHP_MEMMAP_ON_MEMORY
512 depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP
513 depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
515 endif # MEMORY_HOTPLUG
517 # Heavily threaded applications may benefit from splitting the mm-wide
518 # page_table_lock, so that faults on different parts of the user address
519 # space can be handled with less contention: split it at this NR_CPUS.
520 # Default to 4 for wider testing, though 8 might be more appropriate.
521 # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
522 # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
523 # SPARC32 allocates multiple pte tables within a single page, and therefore
524 # a per-page lock leads to problems when multiple tables need to be locked
525 # at the same time (e.g. copy_page_range()).
526 # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page.
528 config SPLIT_PTLOCK_CPUS
530 default "999999" if !MMU
531 default "999999" if ARM && !CPU_CACHE_VIPT
532 default "999999" if PARISC && !PA20
533 default "999999" if SPARC32
536 config ARCH_ENABLE_SPLIT_PMD_PTLOCK
540 # support for memory balloon
541 config MEMORY_BALLOON
545 # support for memory balloon compaction
546 config BALLOON_COMPACTION
547 bool "Allow for balloon memory compaction/migration"
549 depends on COMPACTION && MEMORY_BALLOON
551 Memory fragmentation introduced by ballooning might reduce
552 significantly the number of 2MB contiguous memory blocks that can be
553 used within a guest, thus imposing performance penalties associated
554 with the reduced number of transparent huge pages that could be used
555 by the guest workload. Allowing the compaction & migration for memory
556 pages enlisted as being part of memory balloon devices avoids the
557 scenario aforementioned and helps improving memory defragmentation.
560 # support for memory compaction
562 bool "Allow for memory compaction"
567 Compaction is the only memory management component to form
568 high order (larger physically contiguous) memory blocks
569 reliably. The page allocator relies on compaction heavily and
570 the lack of the feature can lead to unexpected OOM killer
571 invocations for high order memory requests. You shouldn't
572 disable this option unless there really is a strong reason for
573 it and then we would be really interested to hear about that at
577 # support for free page reporting
578 config PAGE_REPORTING
579 bool "Free page reporting"
582 Free page reporting allows for the incremental acquisition of
583 free pages from the buddy allocator for the purpose of reporting
584 those pages to another entity, such as a hypervisor, so that the
585 memory can be freed within the host for other uses.
588 # support for page migration
591 bool "Page migration"
593 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU
595 Allows the migration of the physical location of pages of processes
596 while the virtual addresses are not changed. This is useful in
597 two situations. The first is on NUMA systems to put pages nearer
598 to the processors accessing. The second is when allocating huge
599 pages as migration can relocate pages to satisfy a huge page
600 allocation instead of reclaiming.
602 config DEVICE_MIGRATION
603 def_bool MIGRATION && ZONE_DEVICE
605 config ARCH_ENABLE_HUGEPAGE_MIGRATION
608 config ARCH_ENABLE_THP_MIGRATION
611 config HUGETLB_PAGE_SIZE_VARIABLE
614 Allows the pageblock_order value to be dynamic instead of just standard
615 HUGETLB_PAGE_ORDER when there are multiple HugeTLB page sizes available
618 Note that the pageblock_order cannot exceed MAX_ORDER - 1 and will be
619 clamped down to MAX_ORDER - 1.
622 def_bool (MEMORY_ISOLATION && COMPACTION) || CMA
624 config PHYS_ADDR_T_64BIT
628 bool "Enable bounce buffers"
630 depends on BLOCK && MMU && HIGHMEM
632 Enable bounce buffers for devices that cannot access the full range of
633 memory available to the CPU. Enabled by default when HIGHMEM is
634 selected, but you may say n to override this.
642 bool "Enable KSM for page merging"
646 Enable Kernel Samepage Merging: KSM periodically scans those areas
647 of an application's address space that an app has advised may be
648 mergeable. When it finds pages of identical content, it replaces
649 the many instances by a single page with that content, so
650 saving memory until one or another app needs to modify the content.
651 Recommended for use with KVM, or with other duplicative applications.
652 See Documentation/mm/ksm.rst for more information: KSM is inactive
653 until a program has madvised that an area is MADV_MERGEABLE, and
654 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
656 config DEFAULT_MMAP_MIN_ADDR
657 int "Low address space to protect from user allocation"
661 This is the portion of low virtual memory which should be protected
662 from userspace allocation. Keeping a user from writing to low pages
663 can help reduce the impact of kernel NULL pointer bugs.
665 For most ia64, ppc64 and x86 users with lots of address space
666 a value of 65536 is reasonable and should cause no problems.
667 On arm and other archs it should not be higher than 32768.
668 Programs which use vm86 functionality or have some need to map
669 this low address space will need CAP_SYS_RAWIO or disable this
670 protection by setting the value to 0.
672 This value can be changed after boot using the
673 /proc/sys/vm/mmap_min_addr tunable.
675 config ARCH_SUPPORTS_MEMORY_FAILURE
678 config MEMORY_FAILURE
680 depends on ARCH_SUPPORTS_MEMORY_FAILURE
681 bool "Enable recovery from hardware memory errors"
682 select MEMORY_ISOLATION
685 Enables code to recover from some memory failures on systems
686 with MCA recovery. This allows a system to continue running
687 even when some of its memory has uncorrected errors. This requires
688 special hardware support and typically ECC memory.
690 config HWPOISON_INJECT
691 tristate "HWPoison pages injector"
692 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS
693 select PROC_PAGE_MONITOR
695 config NOMMU_INITIAL_TRIM_EXCESS
696 int "Turn on mmap() excess space trimming before booting"
700 The NOMMU mmap() frequently needs to allocate large contiguous chunks
701 of memory on which to store mappings, but it can only ask the system
702 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
703 more than it requires. To deal with this, mmap() is able to trim off
704 the excess and return it to the allocator.
706 If trimming is enabled, the excess is trimmed off and returned to the
707 system allocator, which can cause extra fragmentation, particularly
708 if there are a lot of transient processes.
710 If trimming is disabled, the excess is kept, but not used, which for
711 long-term mappings means that the space is wasted.
713 Trimming can be dynamically controlled through a sysctl option
714 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of
715 excess pages there must be before trimming should occur, or zero if
716 no trimming is to occur.
718 This option specifies the initial value of this option. The default
719 of 1 says that all excess pages should be trimmed.
721 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
723 config ARCH_WANT_GENERAL_HUGETLB
726 config ARCH_WANTS_THP_SWAP
729 menuconfig TRANSPARENT_HUGEPAGE
730 bool "Transparent Hugepage Support"
731 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE && !PREEMPT_RT
735 Transparent Hugepages allows the kernel to use huge pages and
736 huge tlb transparently to the applications whenever possible.
737 This feature can improve computing performance to certain
738 applications by speeding up page faults during memory
739 allocation, by reducing the number of tlb misses and by speeding
740 up the pagetable walking.
742 If memory constrained on embedded, you may want to say N.
744 if TRANSPARENT_HUGEPAGE
747 prompt "Transparent Hugepage Support sysfs defaults"
748 depends on TRANSPARENT_HUGEPAGE
749 default TRANSPARENT_HUGEPAGE_ALWAYS
751 Selects the sysfs defaults for Transparent Hugepage Support.
753 config TRANSPARENT_HUGEPAGE_ALWAYS
756 Enabling Transparent Hugepage always, can increase the
757 memory footprint of applications without a guaranteed
758 benefit but it will work automatically for all applications.
760 config TRANSPARENT_HUGEPAGE_MADVISE
763 Enabling Transparent Hugepage madvise, will only provide a
764 performance improvement benefit to the applications using
765 madvise(MADV_HUGEPAGE) but it won't risk to increase the
766 memory footprint of applications without a guaranteed
772 depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP
774 Swap transparent huge pages in one piece, without splitting.
775 XXX: For now, swap cluster backing transparent huge page
776 will be split after swapout.
778 For selection by architectures with reasonable THP sizes.
780 config READ_ONLY_THP_FOR_FS
781 bool "Read-only THP for filesystems (EXPERIMENTAL)"
782 depends on TRANSPARENT_HUGEPAGE && SHMEM
785 Allow khugepaged to put read-only file-backed pages in THP.
787 This is marked experimental because it is a new feature. Write
788 support of file THPs will be developed in the next few release
791 endif # TRANSPARENT_HUGEPAGE
794 # UP and nommu archs use km based percpu allocator
796 config NEED_PER_CPU_KM
797 depends on !SMP || !MMU
801 config NEED_PER_CPU_EMBED_FIRST_CHUNK
804 config NEED_PER_CPU_PAGE_FIRST_CHUNK
807 config USE_PERCPU_NUMA_NODE_ID
810 config HAVE_SETUP_PER_CPU_AREA
817 bool "Contiguous Memory Allocator"
820 select MEMORY_ISOLATION
822 This enables the Contiguous Memory Allocator which allows other
823 subsystems to allocate big physically-contiguous blocks of memory.
824 CMA reserves a region of memory and allows only movable pages to
825 be allocated from it. This way, the kernel can use the memory for
826 pagecache and when a subsystem requests for contiguous area, the
827 allocated pages are migrated away to serve the contiguous request.
832 bool "CMA debug messages (DEVELOPMENT)"
833 depends on DEBUG_KERNEL && CMA
835 Turns on debug messages in CMA. This produces KERN_DEBUG
836 messages for every CMA call as well as various messages while
837 processing calls such as dma_alloc_from_contiguous().
838 This option does not affect warning and error messages.
841 bool "CMA debugfs interface"
842 depends on CMA && DEBUG_FS
844 Turns on the DebugFS interface for CMA.
847 bool "CMA information through sysfs interface"
848 depends on CMA && SYSFS
850 This option exposes some sysfs attributes to get information
854 int "Maximum count of the CMA areas"
859 CMA allows to create CMA areas for particular purpose, mainly,
860 used as device private area. This parameter sets the maximum
861 number of CMA area in the system.
863 If unsure, leave the default value "7" in UMA and "19" in NUMA.
865 config MEM_SOFT_DIRTY
866 bool "Track memory changes"
867 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS
868 select PROC_PAGE_MONITOR
870 This option enables memory changes tracking by introducing a
871 soft-dirty bit on pte-s. This bit it set when someone writes
872 into a page just as regular dirty bit, but unlike the latter
873 it can be cleared by hands.
875 See Documentation/admin-guide/mm/soft-dirty.rst for more details.
877 config GENERIC_EARLY_IOREMAP
880 config STACK_MAX_DEFAULT_SIZE_MB
881 int "Default maximum user stack size for 32-bit processes (MB)"
884 depends on STACK_GROWSUP && (!64BIT || COMPAT)
886 This is the maximum stack size in Megabytes in the VM layout of 32-bit
887 user processes when the stack grows upwards (currently only on parisc
888 arch) when the RLIMIT_STACK hard limit is unlimited.
890 A sane initial value is 100 MB.
892 config DEFERRED_STRUCT_PAGE_INIT
893 bool "Defer initialisation of struct pages to kthreads"
895 depends on !NEED_PER_CPU_KM
899 Ordinarily all struct pages are initialised during early boot in a
900 single thread. On very large machines this can take a considerable
901 amount of time. If this option is set, large machines will bring up
902 a subset of memmap at boot and then initialise the rest in parallel.
903 This has a potential performance impact on tasks running early in the
904 lifetime of the system until these kthreads finish the
907 config PAGE_IDLE_FLAG
909 select PAGE_EXTENSION if !64BIT
911 This adds PG_idle and PG_young flags to 'struct page'. PTE Accessed
912 bit writers can set the state of the bit in the flags so that PTE
913 Accessed bit readers may avoid disturbance.
915 config IDLE_PAGE_TRACKING
916 bool "Enable idle page tracking"
917 depends on SYSFS && MMU
918 select PAGE_IDLE_FLAG
920 This feature allows to estimate the amount of user pages that have
921 not been touched during a given period of time. This information can
922 be useful to tune memory cgroup limits and/or for job placement
923 within a compute cluster.
925 See Documentation/admin-guide/mm/idle_page_tracking.rst for
928 config ARCH_HAS_CACHE_LINE_SIZE
931 config ARCH_HAS_CURRENT_STACK_POINTER
934 In support of HARDENED_USERCOPY performing stack variable lifetime
935 checking, an architecture-agnostic way to find the stack pointer
936 is needed. Once an architecture defines an unsigned long global
937 register alias named "current_stack_pointer", this config can be
940 config ARCH_HAS_PTE_DEVMAP
943 config ARCH_HAS_ZONE_DMA_SET
947 bool "Support DMA zone" if ARCH_HAS_ZONE_DMA_SET
948 default y if ARM64 || X86
951 bool "Support DMA32 zone" if ARCH_HAS_ZONE_DMA_SET
956 bool "Device memory (pmem, HMM, etc...) hotplug support"
957 depends on MEMORY_HOTPLUG
958 depends on MEMORY_HOTREMOVE
959 depends on SPARSEMEM_VMEMMAP
960 depends on ARCH_HAS_PTE_DEVMAP
964 Device memory hotplug support allows for establishing pmem,
965 or other device driver discovered memory regions, in the
966 memmap. This allows pfn_to_page() lookups of otherwise
967 "device-physical" addresses which is needed for using a DAX
968 mapping in an O_DIRECT operation, among other things.
970 If FS_DAX is enabled, then say Y.
973 # Helpers to mirror range of the CPU page tables of a process into device page
980 config GET_FREE_REGION
984 config DEVICE_PRIVATE
985 bool "Unaddressable device memory (GPU memory, ...)"
986 depends on ZONE_DEVICE
987 select GET_FREE_REGION
990 Allows creation of struct pages to represent unaddressable device
991 memory; i.e., memory that is only accessible from the device (or
992 group of devices). You likely also want to select HMM_MIRROR.
997 config ARCH_USES_HIGH_VMA_FLAGS
999 config ARCH_HAS_PKEYS
1002 config VM_EVENT_COUNTERS
1004 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1006 VM event counters are needed for event counts to be shown.
1007 This option allows the disabling of the VM event counters
1008 on EXPERT systems. /proc/vmstat will only show page counts
1009 if VM event counters are disabled.
1012 bool "Collect percpu memory statistics"
1014 This feature collects and exposes statistics via debugfs. The
1015 information includes global and per chunk statistics, which can
1016 be used to help understand percpu memory usage.
1019 bool "Enable infrastructure for get_user_pages()-related unit tests"
1022 Provides /sys/kernel/debug/gup_test, which in turn provides a way
1023 to make ioctl calls that can launch kernel-based unit tests for
1024 the get_user_pages*() and pin_user_pages*() family of API calls.
1026 These tests include benchmark testing of the _fast variants of
1027 get_user_pages*() and pin_user_pages*(), as well as smoke tests of
1028 the non-_fast variants.
1030 There is also a sub-test that allows running dump_page() on any
1031 of up to eight pages (selected by command line args) within the
1032 range of user-space addresses. These pages are either pinned via
1033 pin_user_pages*(), or pinned via get_user_pages*(), as specified
1034 by other command line arguments.
1036 See tools/testing/selftests/vm/gup_test.c
1038 comment "GUP_TEST needs to have DEBUG_FS enabled"
1039 depends on !GUP_TEST && !DEBUG_FS
1041 config GUP_GET_PTE_LOW_HIGH
1044 config ARCH_HAS_PTE_SPECIAL
1048 # Some architectures require a special hugepage directory format that is
1049 # required to support multiple hugepage sizes. For example a4fe3ce76
1050 # "powerpc/mm: Allow more flexible layouts for hugepage pagetables"
1051 # introduced it on powerpc. This allows for a more flexible hugepage
1052 # pagetable layouts.
1054 config ARCH_HAS_HUGEPD
1057 config MAPPING_DIRTY_HELPERS
1063 config KMAP_LOCAL_NON_LINEAR_PTE_ARRAY
1066 # struct io_mapping based helper. Selected by drivers that need them
1071 def_bool ARCH_HAS_SET_DIRECT_MAP && !EMBEDDED
1073 config ANON_VMA_NAME
1074 bool "Anonymous VMA name support"
1075 depends on PROC_FS && ADVISE_SYSCALLS && MMU
1078 Allow naming anonymous virtual memory areas.
1080 This feature allows assigning names to virtual memory areas. Assigned
1081 names can be later retrieved from /proc/pid/maps and /proc/pid/smaps
1082 and help identifying individual anonymous memory areas.
1083 Assigning a name to anonymous virtual memory area might prevent that
1084 area from being merged with adjacent virtual memory areas due to the
1085 difference in their name.
1088 bool "Enable userfaultfd() system call"
1091 Enable the userfaultfd() system call that allows to intercept and
1092 handle page faults in userland.
1094 config HAVE_ARCH_USERFAULTFD_WP
1097 Arch has userfaultfd write protection support
1099 config HAVE_ARCH_USERFAULTFD_MINOR
1102 Arch has userfaultfd minor fault support
1108 Allows to create marker PTEs for file-backed memory.
1110 config PTE_MARKER_UFFD_WP
1111 bool "Userfaultfd write protection support for shmem/hugetlbfs"
1113 depends on HAVE_ARCH_USERFAULTFD_WP
1117 Allows to create marker PTEs for userfaultfd write protection
1118 purposes. It is required to enable userfaultfd write protection on
1119 file-backed memory types like shmem and hugetlbfs.
1123 bool "Multi-Gen LRU"
1125 # make sure folio->flags has enough spare bits
1126 depends on 64BIT || !SPARSEMEM || SPARSEMEM_VMEMMAP
1128 A high performance LRU implementation to overcommit memory. See
1129 Documentation/admin-guide/mm/multigen_lru.rst for details.
1131 config LRU_GEN_ENABLED
1132 bool "Enable by default"
1135 This option enables the multi-gen LRU by default.
1137 config LRU_GEN_STATS
1138 bool "Full stats for debugging"
1141 Do not enable this option unless you plan to look at historical stats
1142 from evicted generations for debugging purpose.
1144 This option has a per-memcg and per-node memory overhead.
1147 source "mm/damon/Kconfig"