1 config SELECT_MEMORY_MODEL
3 depends on ARCH_SELECT_MEMORY_MODEL
7 depends on SELECT_MEMORY_MODEL
8 default DISCONTIGMEM_MANUAL if ARCH_DISCONTIGMEM_DEFAULT
9 default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT
10 default FLATMEM_MANUAL
14 depends on !(ARCH_DISCONTIGMEM_ENABLE || ARCH_SPARSEMEM_ENABLE) || ARCH_FLATMEM_ENABLE
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: FLATMEM. This is normal
21 Some users of more advanced features like NUMA and
22 memory hotplug may have different options here.
23 DISCONTIGMEM is a more mature, better tested system,
24 but is incompatible with memory hotplug and may suffer
25 decreased performance over SPARSEMEM. If unsure between
26 "Sparse Memory" and "Discontiguous Memory", choose
27 "Discontiguous Memory".
29 If unsure, choose this option (Flat Memory) over any other.
31 config DISCONTIGMEM_MANUAL
32 bool "Discontiguous Memory"
33 depends on ARCH_DISCONTIGMEM_ENABLE
35 This option provides enhanced support for discontiguous
36 memory systems, over FLATMEM. These systems have holes
37 in their physical address spaces, and this option provides
38 more efficient handling of these holes. However, the vast
39 majority of hardware has quite flat address spaces, and
40 can have degraded performance from the extra overhead that
43 Many NUMA configurations will have this as the only option.
45 If unsure, choose "Flat Memory" over this option.
47 config SPARSEMEM_MANUAL
49 depends on ARCH_SPARSEMEM_ENABLE
51 This will be the only option for some systems, including
52 memory hotplug systems. This is normal.
54 For many other systems, this will be an alternative to
55 "Discontiguous Memory". This option provides some potential
56 performance benefits, along with decreased code complexity,
57 but it is newer, and more experimental.
59 If unsure, choose "Discontiguous Memory" or "Flat Memory"
66 depends on (!SELECT_MEMORY_MODEL && ARCH_DISCONTIGMEM_ENABLE) || DISCONTIGMEM_MANUAL
70 depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL
74 depends on (!DISCONTIGMEM && !SPARSEMEM) || FLATMEM_MANUAL
76 config FLAT_NODE_MEM_MAP
81 # Both the NUMA code and DISCONTIGMEM use arrays of pg_data_t's
82 # to represent different areas of memory. This variable allows
83 # those dependencies to exist individually.
85 config NEED_MULTIPLE_NODES
87 depends on DISCONTIGMEM || NUMA
89 config HAVE_MEMORY_PRESENT
91 depends on ARCH_HAVE_MEMORY_PRESENT || SPARSEMEM
94 # SPARSEMEM_EXTREME (which is the default) does some bootmem
95 # allocations when memory_present() is called. If this cannot
96 # be done on your architecture, select this option. However,
97 # statically allocating the mem_section[] array can potentially
98 # consume vast quantities of .bss, so be careful.
100 # This option will also potentially produce smaller runtime code
101 # with gcc 3.4 and later.
103 config SPARSEMEM_STATIC
107 # Architecture platforms which require a two level mem_section in SPARSEMEM
108 # must select this option. This is usually for architecture platforms with
109 # an extremely sparse physical address space.
111 config SPARSEMEM_EXTREME
113 depends on SPARSEMEM && !SPARSEMEM_STATIC
115 config SPARSEMEM_VMEMMAP_ENABLE
118 config SPARSEMEM_ALLOC_MEM_MAP_TOGETHER
120 depends on SPARSEMEM && X86_64
122 config SPARSEMEM_VMEMMAP
123 bool "Sparse Memory virtual memmap"
124 depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE
127 SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise
128 pfn_to_page and page_to_pfn operations. This is the most
129 efficient option when sufficient kernel resources are available.
134 config HAVE_MEMBLOCK_NODE_MAP
137 config HAVE_MEMBLOCK_PHYS_MAP
140 config HAVE_GENERIC_RCU_GUP
143 config ARCH_DISCARD_MEMBLOCK
149 config MEMORY_ISOLATION
153 bool "Enable to assign a node which has only movable memory"
154 depends on HAVE_MEMBLOCK
155 depends on NO_BOOTMEM
160 Allow a node to have only movable memory. Pages used by the kernel,
161 such as direct mapping pages cannot be migrated. So the corresponding
162 memory device cannot be hotplugged. This option allows the following
164 - When the system is booting, node full of hotpluggable memory can
165 be arranged to have only movable memory so that the whole node can
166 be hot-removed. (need movable_node boot option specified).
167 - After the system is up, the option allows users to online all the
168 memory of a node as movable memory so that the whole node can be
171 Users who don't use the memory hotplug feature are fine with this
172 option on since they don't specify movable_node boot option or they
173 don't online memory as movable.
175 Say Y here if you want to hotplug a whole node.
176 Say N here if you want kernel to use memory on all nodes evenly.
179 # Only be set on architectures that have completely implemented memory hotplug
180 # feature. If you are not sure, don't touch it.
182 config HAVE_BOOTMEM_INFO_NODE
185 # eventually, we can have this option just 'select SPARSEMEM'
186 config MEMORY_HOTPLUG
187 bool "Allow for memory hot-add"
188 depends on SPARSEMEM || X86_64_ACPI_NUMA
189 depends on ARCH_ENABLE_MEMORY_HOTPLUG
190 depends on (IA64 || X86 || PPC_BOOK3S_64 || SUPERH || S390)
192 config MEMORY_HOTPLUG_SPARSE
194 depends on SPARSEMEM && MEMORY_HOTPLUG
196 config MEMORY_HOTREMOVE
197 bool "Allow for memory hot remove"
198 select MEMORY_ISOLATION
199 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64)
200 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
203 # Heavily threaded applications may benefit from splitting the mm-wide
204 # page_table_lock, so that faults on different parts of the user address
205 # space can be handled with less contention: split it at this NR_CPUS.
206 # Default to 4 for wider testing, though 8 might be more appropriate.
207 # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
208 # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
209 # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page.
211 config SPLIT_PTLOCK_CPUS
213 default "999999" if !MMU
214 default "999999" if ARM && !CPU_CACHE_VIPT
215 default "999999" if PARISC && !PA20
218 config ARCH_ENABLE_SPLIT_PMD_PTLOCK
222 # support for memory balloon
223 config MEMORY_BALLOON
227 # support for memory balloon compaction
228 config BALLOON_COMPACTION
229 bool "Allow for balloon memory compaction/migration"
231 depends on COMPACTION && MEMORY_BALLOON
233 Memory fragmentation introduced by ballooning might reduce
234 significantly the number of 2MB contiguous memory blocks that can be
235 used within a guest, thus imposing performance penalties associated
236 with the reduced number of transparent huge pages that could be used
237 by the guest workload. Allowing the compaction & migration for memory
238 pages enlisted as being part of memory balloon devices avoids the
239 scenario aforementioned and helps improving memory defragmentation.
242 # support for memory compaction
244 bool "Allow for memory compaction"
249 Allows the compaction of memory for the allocation of huge pages.
252 # support for page migration
255 bool "Page migration"
257 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU
259 Allows the migration of the physical location of pages of processes
260 while the virtual addresses are not changed. This is useful in
261 two situations. The first is on NUMA systems to put pages nearer
262 to the processors accessing. The second is when allocating huge
263 pages as migration can relocate pages to satisfy a huge page
264 allocation instead of reclaiming.
266 config ARCH_ENABLE_HUGEPAGE_MIGRATION
269 config PHYS_ADDR_T_64BIT
270 def_bool 64BIT || ARCH_PHYS_ADDR_T_64BIT
274 default "0" if !ZONE_DMA
278 bool "Enable bounce buffers"
280 depends on BLOCK && MMU && (ZONE_DMA || HIGHMEM)
282 Enable bounce buffers for devices that cannot access
283 the full range of memory available to the CPU. Enabled
284 by default when ZONE_DMA or HIGHMEM is selected, but you
285 may say n to override this.
287 # On the 'tile' arch, USB OHCI needs the bounce pool since tilegx will often
288 # have more than 4GB of memory, but we don't currently use the IOTLB to present
289 # a 32-bit address to OHCI. So we need to use a bounce pool instead.
290 config NEED_BOUNCE_POOL
292 default y if TILE && USB_OHCI_HCD
303 An architecture should select this if it implements the
304 deprecated interface virt_to_bus(). All new architectures
305 should probably not select this.
313 bool "Enable KSM for page merging"
316 Enable Kernel Samepage Merging: KSM periodically scans those areas
317 of an application's address space that an app has advised may be
318 mergeable. When it finds pages of identical content, it replaces
319 the many instances by a single page with that content, so
320 saving memory until one or another app needs to modify the content.
321 Recommended for use with KVM, or with other duplicative applications.
322 See Documentation/vm/ksm.txt for more information: KSM is inactive
323 until a program has madvised that an area is MADV_MERGEABLE, and
324 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
326 config DEFAULT_MMAP_MIN_ADDR
327 int "Low address space to protect from user allocation"
331 This is the portion of low virtual memory which should be protected
332 from userspace allocation. Keeping a user from writing to low pages
333 can help reduce the impact of kernel NULL pointer bugs.
335 For most ia64, ppc64 and x86 users with lots of address space
336 a value of 65536 is reasonable and should cause no problems.
337 On arm and other archs it should not be higher than 32768.
338 Programs which use vm86 functionality or have some need to map
339 this low address space will need CAP_SYS_RAWIO or disable this
340 protection by setting the value to 0.
342 This value can be changed after boot using the
343 /proc/sys/vm/mmap_min_addr tunable.
345 config ARCH_SUPPORTS_MEMORY_FAILURE
348 config MEMORY_FAILURE
350 depends on ARCH_SUPPORTS_MEMORY_FAILURE
351 bool "Enable recovery from hardware memory errors"
352 select MEMORY_ISOLATION
355 Enables code to recover from some memory failures on systems
356 with MCA recovery. This allows a system to continue running
357 even when some of its memory has uncorrected errors. This requires
358 special hardware support and typically ECC memory.
360 config HWPOISON_INJECT
361 tristate "HWPoison pages injector"
362 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS
363 select PROC_PAGE_MONITOR
365 config NOMMU_INITIAL_TRIM_EXCESS
366 int "Turn on mmap() excess space trimming before booting"
370 The NOMMU mmap() frequently needs to allocate large contiguous chunks
371 of memory on which to store mappings, but it can only ask the system
372 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
373 more than it requires. To deal with this, mmap() is able to trim off
374 the excess and return it to the allocator.
376 If trimming is enabled, the excess is trimmed off and returned to the
377 system allocator, which can cause extra fragmentation, particularly
378 if there are a lot of transient processes.
380 If trimming is disabled, the excess is kept, but not used, which for
381 long-term mappings means that the space is wasted.
383 Trimming can be dynamically controlled through a sysctl option
384 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of
385 excess pages there must be before trimming should occur, or zero if
386 no trimming is to occur.
388 This option specifies the initial value of this option. The default
389 of 1 says that all excess pages should be trimmed.
391 See Documentation/nommu-mmap.txt for more information.
393 config TRANSPARENT_HUGEPAGE
394 bool "Transparent Hugepage Support"
395 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE
398 Transparent Hugepages allows the kernel to use huge pages and
399 huge tlb transparently to the applications whenever possible.
400 This feature can improve computing performance to certain
401 applications by speeding up page faults during memory
402 allocation, by reducing the number of tlb misses and by speeding
403 up the pagetable walking.
405 If memory constrained on embedded, you may want to say N.
408 prompt "Transparent Hugepage Support sysfs defaults"
409 depends on TRANSPARENT_HUGEPAGE
410 default TRANSPARENT_HUGEPAGE_ALWAYS
412 Selects the sysfs defaults for Transparent Hugepage Support.
414 config TRANSPARENT_HUGEPAGE_ALWAYS
417 Enabling Transparent Hugepage always, can increase the
418 memory footprint of applications without a guaranteed
419 benefit but it will work automatically for all applications.
421 config TRANSPARENT_HUGEPAGE_MADVISE
424 Enabling Transparent Hugepage madvise, will only provide a
425 performance improvement benefit to the applications using
426 madvise(MADV_HUGEPAGE) but it won't risk to increase the
427 memory footprint of applications without a guaranteed
432 # UP and nommu archs use km based percpu allocator
434 config NEED_PER_CPU_KM
440 bool "Enable cleancache driver to cache clean pages if tmem is present"
443 Cleancache can be thought of as a page-granularity victim cache
444 for clean pages that the kernel's pageframe replacement algorithm
445 (PFRA) would like to keep around, but can't since there isn't enough
446 memory. So when the PFRA "evicts" a page, it first attempts to use
447 cleancache code to put the data contained in that page into
448 "transcendent memory", memory that is not directly accessible or
449 addressable by the kernel and is of unknown and possibly
450 time-varying size. And when a cleancache-enabled
451 filesystem wishes to access a page in a file on disk, it first
452 checks cleancache to see if it already contains it; if it does,
453 the page is copied into the kernel and a disk access is avoided.
454 When a transcendent memory driver is available (such as zcache or
455 Xen transcendent memory), a significant I/O reduction
456 may be achieved. When none is available, all cleancache calls
457 are reduced to a single pointer-compare-against-NULL resulting
458 in a negligible performance hit.
460 If unsure, say Y to enable cleancache
463 bool "Enable frontswap to cache swap pages if tmem is present"
467 Frontswap is so named because it can be thought of as the opposite
468 of a "backing" store for a swap device. The data is stored into
469 "transcendent memory", memory that is not directly accessible or
470 addressable by the kernel and is of unknown and possibly
471 time-varying size. When space in transcendent memory is available,
472 a significant swap I/O reduction may be achieved. When none is
473 available, all frontswap calls are reduced to a single pointer-
474 compare-against-NULL resulting in a negligible performance hit
475 and swap data is stored as normal on the matching swap device.
477 If unsure, say Y to enable frontswap.
480 bool "Contiguous Memory Allocator"
481 depends on HAVE_MEMBLOCK && MMU
483 select MEMORY_ISOLATION
485 This enables the Contiguous Memory Allocator which allows other
486 subsystems to allocate big physically-contiguous blocks of memory.
487 CMA reserves a region of memory and allows only movable pages to
488 be allocated from it. This way, the kernel can use the memory for
489 pagecache and when a subsystem requests for contiguous area, the
490 allocated pages are migrated away to serve the contiguous request.
495 bool "CMA debug messages (DEVELOPMENT)"
496 depends on DEBUG_KERNEL && CMA
498 Turns on debug messages in CMA. This produces KERN_DEBUG
499 messages for every CMA call as well as various messages while
500 processing calls such as dma_alloc_from_contiguous().
501 This option does not affect warning and error messages.
504 bool "CMA debugfs interface"
505 depends on CMA && DEBUG_FS
507 Turns on the DebugFS interface for CMA.
510 int "Maximum count of the CMA areas"
514 CMA allows to create CMA areas for particular purpose, mainly,
515 used as device private area. This parameter sets the maximum
516 number of CMA area in the system.
518 If unsure, leave the default value "7".
520 config MEM_SOFT_DIRTY
521 bool "Track memory changes"
522 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS
523 select PROC_PAGE_MONITOR
525 This option enables memory changes tracking by introducing a
526 soft-dirty bit on pte-s. This bit it set when someone writes
527 into a page just as regular dirty bit, but unlike the latter
528 it can be cleared by hands.
530 See Documentation/vm/soft-dirty.txt for more details.
533 bool "Compressed cache for swap pages (EXPERIMENTAL)"
534 depends on FRONTSWAP && CRYPTO=y
539 A lightweight compressed cache for swap pages. It takes
540 pages that are in the process of being swapped out and attempts to
541 compress them into a dynamically allocated RAM-based memory pool.
542 This can result in a significant I/O reduction on swap device and,
543 in the case where decompressing from RAM is faster that swap device
544 reads, can also improve workload performance.
546 This is marked experimental because it is a new feature (as of
547 v3.11) that interacts heavily with memory reclaim. While these
548 interactions don't cause any known issues on simple memory setups,
549 they have not be fully explored on the large set of potential
550 configurations and workloads that exist.
553 tristate "Common API for compressed memory storage"
556 Compressed memory storage API. This allows using either zbud or
560 tristate "Low density storage for compressed pages"
563 A special purpose allocator for storing compressed pages.
564 It is designed to store up to two compressed pages per physical
565 page. While this design limits storage density, it has simple and
566 deterministic reclaim properties that make it preferable to a higher
567 density approach when reclaim will be used.
570 tristate "Memory allocator for compressed pages"
574 zsmalloc is a slab-based memory allocator designed to store
575 compressed RAM pages. zsmalloc uses virtual memory mapping
576 in order to reduce fragmentation. However, this results in a
577 non-standard allocator interface where a handle, not a pointer, is
578 returned by an alloc(). This handle must be mapped in order to
579 access the allocated space.
581 config PGTABLE_MAPPING
582 bool "Use page table mapping to access object in zsmalloc"
585 By default, zsmalloc uses a copy-based object mapping method to
586 access allocations that span two pages. However, if a particular
587 architecture (ex, ARM) performs VM mapping faster than copying,
588 then you should select this. This causes zsmalloc to use page table
589 mapping rather than copying for object mapping.
591 You can check speed with zsmalloc benchmark:
592 https://github.com/spartacus06/zsmapbench
595 bool "Export zsmalloc statistics"
599 This option enables code in the zsmalloc to collect various
600 statistics about whats happening in zsmalloc and exports that
601 information to userspace via debugfs.
604 config GENERIC_EARLY_IOREMAP
607 config MAX_STACK_SIZE_MB
608 int "Maximum user stack size for 32-bit processes (MB)"
612 depends on STACK_GROWSUP && (!64BIT || COMPAT)
614 This is the maximum stack size in Megabytes in the VM layout of 32-bit
615 user processes when the stack grows upwards (currently only on parisc
616 and metag arch). The stack will be located at the highest memory
617 address minus the given value, unless the RLIMIT_STACK hard limit is
618 changed to a smaller value in which case that is used.
620 A sane initial value is 80 MB.
622 # For architectures that support deferred memory initialisation
623 config ARCH_SUPPORTS_DEFERRED_STRUCT_PAGE_INIT
626 config DEFERRED_STRUCT_PAGE_INIT
627 bool "Defer initialisation of struct pages to kswapd"
629 depends on ARCH_SUPPORTS_DEFERRED_STRUCT_PAGE_INIT
630 depends on MEMORY_HOTPLUG
632 Ordinarily all struct pages are initialised during early boot in a
633 single thread. On very large machines this can take a considerable
634 amount of time. If this option is set, large machines will bring up
635 a subset of memmap at boot and then initialise the rest in parallel
636 when kswapd starts. This has a potential performance impact on
637 processes running early in the lifetime of the systemm until kswapd
638 finishes the initialisation.
640 config IDLE_PAGE_TRACKING
641 bool "Enable idle page tracking"
642 depends on SYSFS && MMU
643 select PAGE_EXTENSION if !64BIT
645 This feature allows to estimate the amount of user pages that have
646 not been touched during a given period of time. This information can
647 be useful to tune memory cgroup limits and/or for job placement
648 within a compute cluster.
650 See Documentation/vm/idle_page_tracking.txt for more details.
653 bool "Device memory (pmem, etc...) hotplug support" if EXPERT
656 depends on MEMORY_HOTPLUG
657 depends on MEMORY_HOTREMOVE
658 depends on X86_64 #arch_add_memory() comprehends device memory
661 Device memory hotplug support allows for establishing pmem,
662 or other device driver discovered memory regions, in the
663 memmap. This allows pfn_to_page() lookups of otherwise
664 "device-physical" addresses which is needed for using a DAX
665 mapping in an O_DIRECT operation, among other things.
667 If FS_DAX is enabled, then say Y.