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
92 # SPARSEMEM_EXTREME (which is the default) does some bootmem
93 # allocations when sparse_init() is called. If this cannot
94 # be done on your architecture, select this option. However,
95 # statically allocating the mem_section[] array can potentially
96 # consume vast quantities of .bss, so be careful.
98 # This option will also potentially produce smaller runtime code
99 # with gcc 3.4 and later.
101 config SPARSEMEM_STATIC
105 # Architecture platforms which require a two level mem_section in SPARSEMEM
106 # must select this option. This is usually for architecture platforms with
107 # an extremely sparse physical address space.
109 config SPARSEMEM_EXTREME
111 depends on SPARSEMEM && !SPARSEMEM_STATIC
113 config SPARSEMEM_VMEMMAP_ENABLE
116 config SPARSEMEM_VMEMMAP
117 bool "Sparse Memory virtual memmap"
118 depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE
121 SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise
122 pfn_to_page and page_to_pfn operations. This is the most
123 efficient option when sufficient kernel resources are available.
125 config HAVE_MEMBLOCK_PHYS_MAP
132 # Don't discard allocated memory used to track "memory" and "reserved" memblocks
133 # after early boot, so it can still be used to test for validity of memory.
134 # Also, memblocks are updated with memory hot(un)plug.
135 config ARCH_KEEP_MEMBLOCK
138 # Keep arch NUMA mapping infrastructure post-init.
139 config NUMA_KEEP_MEMINFO
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 select MEMORY_ISOLATION
156 depends on SPARSEMEM || X86_64_ACPI_NUMA
157 depends on ARCH_ENABLE_MEMORY_HOTPLUG
158 depends on 64BIT || BROKEN
159 select NUMA_KEEP_MEMINFO if NUMA
161 config MEMORY_HOTPLUG_SPARSE
163 depends on SPARSEMEM && MEMORY_HOTPLUG
165 config MEMORY_HOTPLUG_DEFAULT_ONLINE
166 bool "Online the newly added memory blocks by default"
167 depends on MEMORY_HOTPLUG
169 This option sets the default policy setting for memory hotplug
170 onlining policy (/sys/devices/system/memory/auto_online_blocks) which
171 determines what happens to newly added memory regions. Policy setting
172 can always be changed at runtime.
173 See Documentation/admin-guide/mm/memory-hotplug.rst for more information.
175 Say Y here if you want all hot-plugged memory blocks to appear in
176 'online' state by default.
177 Say N here if you want the default policy to keep all hot-plugged
178 memory blocks in 'offline' state.
180 config MEMORY_HOTREMOVE
181 bool "Allow for memory hot remove"
182 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64)
183 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
186 # Heavily threaded applications may benefit from splitting the mm-wide
187 # page_table_lock, so that faults on different parts of the user address
188 # space can be handled with less contention: split it at this NR_CPUS.
189 # Default to 4 for wider testing, though 8 might be more appropriate.
190 # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
191 # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
192 # SPARC32 allocates multiple pte tables within a single page, and therefore
193 # a per-page lock leads to problems when multiple tables need to be locked
194 # at the same time (e.g. copy_page_range()).
195 # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page.
197 config SPLIT_PTLOCK_CPUS
199 default "999999" if !MMU
200 default "999999" if ARM && !CPU_CACHE_VIPT
201 default "999999" if PARISC && !PA20
202 default "999999" if SPARC32
205 config ARCH_ENABLE_SPLIT_PMD_PTLOCK
209 # support for memory balloon
210 config MEMORY_BALLOON
214 # support for memory balloon compaction
215 config BALLOON_COMPACTION
216 bool "Allow for balloon memory compaction/migration"
218 depends on COMPACTION && MEMORY_BALLOON
220 Memory fragmentation introduced by ballooning might reduce
221 significantly the number of 2MB contiguous memory blocks that can be
222 used within a guest, thus imposing performance penalties associated
223 with the reduced number of transparent huge pages that could be used
224 by the guest workload. Allowing the compaction & migration for memory
225 pages enlisted as being part of memory balloon devices avoids the
226 scenario aforementioned and helps improving memory defragmentation.
229 # support for memory compaction
231 bool "Allow for memory compaction"
236 Compaction is the only memory management component to form
237 high order (larger physically contiguous) memory blocks
238 reliably. The page allocator relies on compaction heavily and
239 the lack of the feature can lead to unexpected OOM killer
240 invocations for high order memory requests. You shouldn't
241 disable this option unless there really is a strong reason for
242 it and then we would be really interested to hear about that at
246 # support for free page reporting
247 config PAGE_REPORTING
248 bool "Free page reporting"
251 Free page reporting allows for the incremental acquisition of
252 free pages from the buddy allocator for the purpose of reporting
253 those pages to another entity, such as a hypervisor, so that the
254 memory can be freed within the host for other uses.
257 # support for page migration
260 bool "Page migration"
262 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU
264 Allows the migration of the physical location of pages of processes
265 while the virtual addresses are not changed. This is useful in
266 two situations. The first is on NUMA systems to put pages nearer
267 to the processors accessing. The second is when allocating huge
268 pages as migration can relocate pages to satisfy a huge page
269 allocation instead of reclaiming.
271 config ARCH_ENABLE_HUGEPAGE_MIGRATION
274 config ARCH_ENABLE_THP_MIGRATION
278 def_bool (MEMORY_ISOLATION && COMPACTION) || CMA
280 config PHYS_ADDR_T_64BIT
284 bool "Enable bounce buffers"
286 depends on BLOCK && MMU && HIGHMEM
288 Enable bounce buffers for devices that cannot access the full range of
289 memory available to the CPU. Enabled by default when HIGHMEM is
290 selected, but you may say n to override this.
295 An architecture should select this if it implements the
296 deprecated interface virt_to_bus(). All new architectures
297 should probably not select this.
306 bool "Enable KSM for page merging"
310 Enable Kernel Samepage Merging: KSM periodically scans those areas
311 of an application's address space that an app has advised may be
312 mergeable. When it finds pages of identical content, it replaces
313 the many instances by a single page with that content, so
314 saving memory until one or another app needs to modify the content.
315 Recommended for use with KVM, or with other duplicative applications.
316 See Documentation/vm/ksm.rst for more information: KSM is inactive
317 until a program has madvised that an area is MADV_MERGEABLE, and
318 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
320 config DEFAULT_MMAP_MIN_ADDR
321 int "Low address space to protect from user allocation"
325 This is the portion of low virtual memory which should be protected
326 from userspace allocation. Keeping a user from writing to low pages
327 can help reduce the impact of kernel NULL pointer bugs.
329 For most ia64, ppc64 and x86 users with lots of address space
330 a value of 65536 is reasonable and should cause no problems.
331 On arm and other archs it should not be higher than 32768.
332 Programs which use vm86 functionality or have some need to map
333 this low address space will need CAP_SYS_RAWIO or disable this
334 protection by setting the value to 0.
336 This value can be changed after boot using the
337 /proc/sys/vm/mmap_min_addr tunable.
339 config ARCH_SUPPORTS_MEMORY_FAILURE
342 config MEMORY_FAILURE
344 depends on ARCH_SUPPORTS_MEMORY_FAILURE
345 bool "Enable recovery from hardware memory errors"
346 select MEMORY_ISOLATION
349 Enables code to recover from some memory failures on systems
350 with MCA recovery. This allows a system to continue running
351 even when some of its memory has uncorrected errors. This requires
352 special hardware support and typically ECC memory.
354 config HWPOISON_INJECT
355 tristate "HWPoison pages injector"
356 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS
357 select PROC_PAGE_MONITOR
359 config NOMMU_INITIAL_TRIM_EXCESS
360 int "Turn on mmap() excess space trimming before booting"
364 The NOMMU mmap() frequently needs to allocate large contiguous chunks
365 of memory on which to store mappings, but it can only ask the system
366 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
367 more than it requires. To deal with this, mmap() is able to trim off
368 the excess and return it to the allocator.
370 If trimming is enabled, the excess is trimmed off and returned to the
371 system allocator, which can cause extra fragmentation, particularly
372 if there are a lot of transient processes.
374 If trimming is disabled, the excess is kept, but not used, which for
375 long-term mappings means that the space is wasted.
377 Trimming can be dynamically controlled through a sysctl option
378 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of
379 excess pages there must be before trimming should occur, or zero if
380 no trimming is to occur.
382 This option specifies the initial value of this option. The default
383 of 1 says that all excess pages should be trimmed.
385 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
387 config TRANSPARENT_HUGEPAGE
388 bool "Transparent Hugepage Support"
389 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE
393 Transparent Hugepages allows the kernel to use huge pages and
394 huge tlb transparently to the applications whenever possible.
395 This feature can improve computing performance to certain
396 applications by speeding up page faults during memory
397 allocation, by reducing the number of tlb misses and by speeding
398 up the pagetable walking.
400 If memory constrained on embedded, you may want to say N.
403 prompt "Transparent Hugepage Support sysfs defaults"
404 depends on TRANSPARENT_HUGEPAGE
405 default TRANSPARENT_HUGEPAGE_ALWAYS
407 Selects the sysfs defaults for Transparent Hugepage Support.
409 config TRANSPARENT_HUGEPAGE_ALWAYS
412 Enabling Transparent Hugepage always, can increase the
413 memory footprint of applications without a guaranteed
414 benefit but it will work automatically for all applications.
416 config TRANSPARENT_HUGEPAGE_MADVISE
419 Enabling Transparent Hugepage madvise, will only provide a
420 performance improvement benefit to the applications using
421 madvise(MADV_HUGEPAGE) but it won't risk to increase the
422 memory footprint of applications without a guaranteed
426 config ARCH_WANTS_THP_SWAP
431 depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP
433 Swap transparent huge pages in one piece, without splitting.
434 XXX: For now, swap cluster backing transparent huge page
435 will be split after swapout.
437 For selection by architectures with reasonable THP sizes.
440 # UP and nommu archs use km based percpu allocator
442 config NEED_PER_CPU_KM
448 bool "Enable cleancache driver to cache clean pages if tmem is present"
450 Cleancache can be thought of as a page-granularity victim cache
451 for clean pages that the kernel's pageframe replacement algorithm
452 (PFRA) would like to keep around, but can't since there isn't enough
453 memory. So when the PFRA "evicts" a page, it first attempts to use
454 cleancache code to put the data contained in that page into
455 "transcendent memory", memory that is not directly accessible or
456 addressable by the kernel and is of unknown and possibly
457 time-varying size. And when a cleancache-enabled
458 filesystem wishes to access a page in a file on disk, it first
459 checks cleancache to see if it already contains it; if it does,
460 the page is copied into the kernel and a disk access is avoided.
461 When a transcendent memory driver is available (such as zcache or
462 Xen transcendent memory), a significant I/O reduction
463 may be achieved. When none is available, all cleancache calls
464 are reduced to a single pointer-compare-against-NULL resulting
465 in a negligible performance hit.
467 If unsure, say Y to enable cleancache
470 bool "Enable frontswap to cache swap pages if tmem is present"
473 Frontswap is so named because it can be thought of as the opposite
474 of a "backing" store for a swap device. The data is stored into
475 "transcendent memory", memory that is not directly accessible or
476 addressable by the kernel and is of unknown and possibly
477 time-varying size. When space in transcendent memory is available,
478 a significant swap I/O reduction may be achieved. When none is
479 available, all frontswap calls are reduced to a single pointer-
480 compare-against-NULL resulting in a negligible performance hit
481 and swap data is stored as normal on the matching swap device.
483 If unsure, say Y to enable frontswap.
486 bool "Contiguous Memory Allocator"
489 select MEMORY_ISOLATION
491 This enables the Contiguous Memory Allocator which allows other
492 subsystems to allocate big physically-contiguous blocks of memory.
493 CMA reserves a region of memory and allows only movable pages to
494 be allocated from it. This way, the kernel can use the memory for
495 pagecache and when a subsystem requests for contiguous area, the
496 allocated pages are migrated away to serve the contiguous request.
501 bool "CMA debug messages (DEVELOPMENT)"
502 depends on DEBUG_KERNEL && CMA
504 Turns on debug messages in CMA. This produces KERN_DEBUG
505 messages for every CMA call as well as various messages while
506 processing calls such as dma_alloc_from_contiguous().
507 This option does not affect warning and error messages.
510 bool "CMA debugfs interface"
511 depends on CMA && DEBUG_FS
513 Turns on the DebugFS interface for CMA.
516 int "Maximum count of the CMA areas"
521 CMA allows to create CMA areas for particular purpose, mainly,
522 used as device private area. This parameter sets the maximum
523 number of CMA area in the system.
525 If unsure, leave the default value "7" in UMA and "19" in NUMA.
527 config MEM_SOFT_DIRTY
528 bool "Track memory changes"
529 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS
530 select PROC_PAGE_MONITOR
532 This option enables memory changes tracking by introducing a
533 soft-dirty bit on pte-s. This bit it set when someone writes
534 into a page just as regular dirty bit, but unlike the latter
535 it can be cleared by hands.
537 See Documentation/admin-guide/mm/soft-dirty.rst for more details.
540 bool "Compressed cache for swap pages (EXPERIMENTAL)"
541 depends on FRONTSWAP && CRYPTO=y
544 A lightweight compressed cache for swap pages. It takes
545 pages that are in the process of being swapped out and attempts to
546 compress them into a dynamically allocated RAM-based memory pool.
547 This can result in a significant I/O reduction on swap device and,
548 in the case where decompressing from RAM is faster that swap device
549 reads, can also improve workload performance.
551 This is marked experimental because it is a new feature (as of
552 v3.11) that interacts heavily with memory reclaim. While these
553 interactions don't cause any known issues on simple memory setups,
554 they have not be fully explored on the large set of potential
555 configurations and workloads that exist.
558 prompt "Compressed cache for swap pages default compressor"
560 default ZSWAP_COMPRESSOR_DEFAULT_LZO
562 Selects the default compression algorithm for the compressed cache
565 For an overview what kind of performance can be expected from
566 a particular compression algorithm please refer to the benchmarks
567 available at the following LWN page:
568 https://lwn.net/Articles/751795/
570 If in doubt, select 'LZO'.
572 The selection made here can be overridden by using the kernel
573 command line 'zswap.compressor=' option.
575 config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
577 select CRYPTO_DEFLATE
579 Use the Deflate algorithm as the default compression algorithm.
581 config ZSWAP_COMPRESSOR_DEFAULT_LZO
585 Use the LZO algorithm as the default compression algorithm.
587 config ZSWAP_COMPRESSOR_DEFAULT_842
591 Use the 842 algorithm as the default compression algorithm.
593 config ZSWAP_COMPRESSOR_DEFAULT_LZ4
597 Use the LZ4 algorithm as the default compression algorithm.
599 config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
603 Use the LZ4HC algorithm as the default compression algorithm.
605 config ZSWAP_COMPRESSOR_DEFAULT_ZSTD
609 Use the zstd algorithm as the default compression algorithm.
612 config ZSWAP_COMPRESSOR_DEFAULT
615 default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
616 default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO
617 default "842" if ZSWAP_COMPRESSOR_DEFAULT_842
618 default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4
619 default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
620 default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD
624 prompt "Compressed cache for swap pages default allocator"
626 default ZSWAP_ZPOOL_DEFAULT_ZBUD
628 Selects the default allocator for the compressed cache for
630 The default is 'zbud' for compatibility, however please do
631 read the description of each of the allocators below before
632 making a right choice.
634 The selection made here can be overridden by using the kernel
635 command line 'zswap.zpool=' option.
637 config ZSWAP_ZPOOL_DEFAULT_ZBUD
641 Use the zbud allocator as the default allocator.
643 config ZSWAP_ZPOOL_DEFAULT_Z3FOLD
647 Use the z3fold allocator as the default allocator.
649 config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
653 Use the zsmalloc allocator as the default allocator.
656 config ZSWAP_ZPOOL_DEFAULT
659 default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD
660 default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD
661 default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
664 config ZSWAP_DEFAULT_ON
665 bool "Enable the compressed cache for swap pages by default"
668 If selected, the compressed cache for swap pages will be enabled
669 at boot, otherwise it will be disabled.
671 The selection made here can be overridden by using the kernel
672 command line 'zswap.enabled=' option.
675 tristate "Common API for compressed memory storage"
677 Compressed memory storage API. This allows using either zbud or
681 tristate "Low (Up to 2x) density storage for compressed pages"
683 A special purpose allocator for storing compressed pages.
684 It is designed to store up to two compressed pages per physical
685 page. While this design limits storage density, it has simple and
686 deterministic reclaim properties that make it preferable to a higher
687 density approach when reclaim will be used.
690 tristate "Up to 3x density storage for compressed pages"
693 A special purpose allocator for storing compressed pages.
694 It is designed to store up to three compressed pages per physical
695 page. It is a ZBUD derivative so the simplicity and determinism are
699 tristate "Memory allocator for compressed pages"
702 zsmalloc is a slab-based memory allocator designed to store
703 compressed RAM pages. zsmalloc uses virtual memory mapping
704 in order to reduce fragmentation. However, this results in a
705 non-standard allocator interface where a handle, not a pointer, is
706 returned by an alloc(). This handle must be mapped in order to
707 access the allocated space.
710 bool "Export zsmalloc statistics"
714 This option enables code in the zsmalloc to collect various
715 statistics about what's happening in zsmalloc and exports that
716 information to userspace via debugfs.
719 config GENERIC_EARLY_IOREMAP
722 config STACK_MAX_DEFAULT_SIZE_MB
723 int "Default maximum user stack size for 32-bit processes (MB)"
726 depends on STACK_GROWSUP && (!64BIT || COMPAT)
728 This is the maximum stack size in Megabytes in the VM layout of 32-bit
729 user processes when the stack grows upwards (currently only on parisc
730 arch) when the RLIMIT_STACK hard limit is unlimited.
732 A sane initial value is 100 MB.
734 config DEFERRED_STRUCT_PAGE_INIT
735 bool "Defer initialisation of struct pages to kthreads"
737 depends on !NEED_PER_CPU_KM
741 Ordinarily all struct pages are initialised during early boot in a
742 single thread. On very large machines this can take a considerable
743 amount of time. If this option is set, large machines will bring up
744 a subset of memmap at boot and then initialise the rest in parallel.
745 This has a potential performance impact on tasks running early in the
746 lifetime of the system until these kthreads finish the
749 config IDLE_PAGE_TRACKING
750 bool "Enable idle page tracking"
751 depends on SYSFS && MMU
752 select PAGE_EXTENSION if !64BIT
754 This feature allows to estimate the amount of user pages that have
755 not been touched during a given period of time. This information can
756 be useful to tune memory cgroup limits and/or for job placement
757 within a compute cluster.
759 See Documentation/admin-guide/mm/idle_page_tracking.rst for
762 config ARCH_HAS_PTE_DEVMAP
766 bool "Device memory (pmem, HMM, etc...) hotplug support"
767 depends on MEMORY_HOTPLUG
768 depends on MEMORY_HOTREMOVE
769 depends on SPARSEMEM_VMEMMAP
770 depends on ARCH_HAS_PTE_DEVMAP
774 Device memory hotplug support allows for establishing pmem,
775 or other device driver discovered memory regions, in the
776 memmap. This allows pfn_to_page() lookups of otherwise
777 "device-physical" addresses which is needed for using a DAX
778 mapping in an O_DIRECT operation, among other things.
780 If FS_DAX is enabled, then say Y.
782 config DEV_PAGEMAP_OPS
786 # Helpers to mirror range of the CPU page tables of a process into device page
793 config DEVICE_PRIVATE
794 bool "Unaddressable device memory (GPU memory, ...)"
795 depends on ZONE_DEVICE
796 select DEV_PAGEMAP_OPS
799 Allows creation of struct pages to represent unaddressable device
800 memory; i.e., memory that is only accessible from the device (or
801 group of devices). You likely also want to select HMM_MIRROR.
806 config ARCH_USES_HIGH_VMA_FLAGS
808 config ARCH_HAS_PKEYS
812 bool "Collect percpu memory statistics"
814 This feature collects and exposes statistics via debugfs. The
815 information includes global and per chunk statistics, which can
816 be used to help understand percpu memory usage.
819 bool "Enable infrastructure for get_user_pages()-related unit tests"
822 Provides /sys/kernel/debug/gup_test, which in turn provides a way
823 to make ioctl calls that can launch kernel-based unit tests for
824 the get_user_pages*() and pin_user_pages*() family of API calls.
826 These tests include benchmark testing of the _fast variants of
827 get_user_pages*() and pin_user_pages*(), as well as smoke tests of
828 the non-_fast variants.
830 There is also a sub-test that allows running dump_page() on any
831 of up to eight pages (selected by command line args) within the
832 range of user-space addresses. These pages are either pinned via
833 pin_user_pages*(), or pinned via get_user_pages*(), as specified
834 by other command line arguments.
836 See tools/testing/selftests/vm/gup_test.c
838 comment "GUP_TEST needs to have DEBUG_FS enabled"
839 depends on !GUP_TEST && !DEBUG_FS
841 config GUP_GET_PTE_LOW_HIGH
844 config READ_ONLY_THP_FOR_FS
845 bool "Read-only THP for filesystems (EXPERIMENTAL)"
846 depends on TRANSPARENT_HUGEPAGE && SHMEM
849 Allow khugepaged to put read-only file-backed pages in THP.
851 This is marked experimental because it is a new feature. Write
852 support of file THPs will be developed in the next few release
855 config ARCH_HAS_PTE_SPECIAL
859 # Some architectures require a special hugepage directory format that is
860 # required to support multiple hugepage sizes. For example a4fe3ce76
861 # "powerpc/mm: Allow more flexible layouts for hugepage pagetables"
862 # introduced it on powerpc. This allows for a more flexible hugepage
865 config ARCH_HAS_HUGEPD
868 config MAPPING_DIRTY_HELPERS