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 ARCH_DISCARD_MEMBLOCK
146 config MEMORY_ISOLATION
150 boolean "Enable to assign a node which has only movable memory"
151 depends on HAVE_MEMBLOCK
152 depends on NO_BOOTMEM
157 Allow a node to have only movable memory. Pages used by the kernel,
158 such as direct mapping pages cannot be migrated. So the corresponding
159 memory device cannot be hotplugged. This option allows the following
161 - When the system is booting, node full of hotpluggable memory can
162 be arranged to have only movable memory so that the whole node can
163 be hot-removed. (need movable_node boot option specified).
164 - After the system is up, the option allows users to online all the
165 memory of a node as movable memory so that the whole node can be
168 Users who don't use the memory hotplug feature are fine with this
169 option on since they don't specify movable_node boot option or they
170 don't online memory as movable.
172 Say Y here if you want to hotplug a whole node.
173 Say N here if you want kernel to use memory on all nodes evenly.
176 # Only be set on architectures that have completely implemented memory hotplug
177 # feature. If you are not sure, don't touch it.
179 config HAVE_BOOTMEM_INFO_NODE
182 # eventually, we can have this option just 'select SPARSEMEM'
183 config MEMORY_HOTPLUG
184 bool "Allow for memory hot-add"
185 depends on SPARSEMEM || X86_64_ACPI_NUMA
186 depends on ARCH_ENABLE_MEMORY_HOTPLUG
187 depends on (IA64 || X86 || PPC_BOOK3S_64 || SUPERH || S390)
189 config MEMORY_HOTPLUG_SPARSE
191 depends on SPARSEMEM && MEMORY_HOTPLUG
193 config MEMORY_HOTREMOVE
194 bool "Allow for memory hot remove"
195 select MEMORY_ISOLATION
196 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64)
197 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
201 # If we have space for more page flags then we can enable additional
202 # optimizations and functionality.
204 # Regular Sparsemem takes page flag bits for the sectionid if it does not
205 # use a virtual memmap. Disable extended page flags for 32 bit platforms
206 # that require the use of a sectionid in the page flags.
208 config PAGEFLAGS_EXTENDED
210 depends on 64BIT || SPARSEMEM_VMEMMAP || !SPARSEMEM
212 # Heavily threaded applications may benefit from splitting the mm-wide
213 # page_table_lock, so that faults on different parts of the user address
214 # space can be handled with less contention: split it at this NR_CPUS.
215 # Default to 4 for wider testing, though 8 might be more appropriate.
216 # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
217 # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
218 # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page.
220 config SPLIT_PTLOCK_CPUS
222 default "999999" if !MMU
223 default "999999" if ARM && !CPU_CACHE_VIPT
224 default "999999" if PARISC && !PA20
227 config ARCH_ENABLE_SPLIT_PMD_PTLOCK
231 # support for memory balloon compaction
232 config BALLOON_COMPACTION
233 bool "Allow for balloon memory compaction/migration"
235 depends on COMPACTION && VIRTIO_BALLOON
237 Memory fragmentation introduced by ballooning might reduce
238 significantly the number of 2MB contiguous memory blocks that can be
239 used within a guest, thus imposing performance penalties associated
240 with the reduced number of transparent huge pages that could be used
241 by the guest workload. Allowing the compaction & migration for memory
242 pages enlisted as being part of memory balloon devices avoids the
243 scenario aforementioned and helps improving memory defragmentation.
246 # support for memory compaction
248 bool "Allow for memory compaction"
253 Allows the compaction of memory for the allocation of huge pages.
256 # support for page migration
259 bool "Page migration"
261 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU
263 Allows the migration of the physical location of pages of processes
264 while the virtual addresses are not changed. This is useful in
265 two situations. The first is on NUMA systems to put pages nearer
266 to the processors accessing. The second is when allocating huge
267 pages as migration can relocate pages to satisfy a huge page
268 allocation instead of reclaiming.
270 config ARCH_ENABLE_HUGEPAGE_MIGRATION
273 config PHYS_ADDR_T_64BIT
274 def_bool 64BIT || ARCH_PHYS_ADDR_T_64BIT
278 default "0" if !ZONE_DMA
282 bool "Enable bounce buffers"
284 depends on BLOCK && MMU && (ZONE_DMA || HIGHMEM)
286 Enable bounce buffers for devices that cannot access
287 the full range of memory available to the CPU. Enabled
288 by default when ZONE_DMA or HIGHMEM is selected, but you
289 may say n to override this.
291 # On the 'tile' arch, USB OHCI needs the bounce pool since tilegx will often
292 # have more than 4GB of memory, but we don't currently use the IOTLB to present
293 # a 32-bit address to OHCI. So we need to use a bounce pool instead.
295 # We also use the bounce pool to provide stable page writes for jbd. jbd
296 # initiates buffer writeback without locking the page or setting PG_writeback,
297 # and fixing that behavior (a second time; jbd2 doesn't have this problem) is
298 # a major rework effort. Instead, use the bounce buffer to snapshot pages
299 # (until jbd goes away). The only jbd user is ext3.
300 config NEED_BOUNCE_POOL
302 default y if (TILE && USB_OHCI_HCD) || (BLK_DEV_INTEGRITY && JBD)
313 An architecture should select this if it implements the
314 deprecated interface virt_to_bus(). All new architectures
315 should probably not select this.
322 bool "Enable KSM for page merging"
325 Enable Kernel Samepage Merging: KSM periodically scans those areas
326 of an application's address space that an app has advised may be
327 mergeable. When it finds pages of identical content, it replaces
328 the many instances by a single page with that content, so
329 saving memory until one or another app needs to modify the content.
330 Recommended for use with KVM, or with other duplicative applications.
331 See Documentation/vm/ksm.txt for more information: KSM is inactive
332 until a program has madvised that an area is MADV_MERGEABLE, and
333 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
335 config DEFAULT_MMAP_MIN_ADDR
336 int "Low address space to protect from user allocation"
340 This is the portion of low virtual memory which should be protected
341 from userspace allocation. Keeping a user from writing to low pages
342 can help reduce the impact of kernel NULL pointer bugs.
344 For most ia64, ppc64 and x86 users with lots of address space
345 a value of 65536 is reasonable and should cause no problems.
346 On arm and other archs it should not be higher than 32768.
347 Programs which use vm86 functionality or have some need to map
348 this low address space will need CAP_SYS_RAWIO or disable this
349 protection by setting the value to 0.
351 This value can be changed after boot using the
352 /proc/sys/vm/mmap_min_addr tunable.
354 config ARCH_SUPPORTS_MEMORY_FAILURE
357 config MEMORY_FAILURE
359 depends on ARCH_SUPPORTS_MEMORY_FAILURE
360 bool "Enable recovery from hardware memory errors"
361 select MEMORY_ISOLATION
363 Enables code to recover from some memory failures on systems
364 with MCA recovery. This allows a system to continue running
365 even when some of its memory has uncorrected errors. This requires
366 special hardware support and typically ECC memory.
368 config HWPOISON_INJECT
369 tristate "HWPoison pages injector"
370 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS
371 select PROC_PAGE_MONITOR
373 config NOMMU_INITIAL_TRIM_EXCESS
374 int "Turn on mmap() excess space trimming before booting"
378 The NOMMU mmap() frequently needs to allocate large contiguous chunks
379 of memory on which to store mappings, but it can only ask the system
380 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
381 more than it requires. To deal with this, mmap() is able to trim off
382 the excess and return it to the allocator.
384 If trimming is enabled, the excess is trimmed off and returned to the
385 system allocator, which can cause extra fragmentation, particularly
386 if there are a lot of transient processes.
388 If trimming is disabled, the excess is kept, but not used, which for
389 long-term mappings means that the space is wasted.
391 Trimming can be dynamically controlled through a sysctl option
392 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of
393 excess pages there must be before trimming should occur, or zero if
394 no trimming is to occur.
396 This option specifies the initial value of this option. The default
397 of 1 says that all excess pages should be trimmed.
399 See Documentation/nommu-mmap.txt for more information.
401 config TRANSPARENT_HUGEPAGE
402 bool "Transparent Hugepage Support"
403 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE
406 Transparent Hugepages allows the kernel to use huge pages and
407 huge tlb transparently to the applications whenever possible.
408 This feature can improve computing performance to certain
409 applications by speeding up page faults during memory
410 allocation, by reducing the number of tlb misses and by speeding
411 up the pagetable walking.
413 If memory constrained on embedded, you may want to say N.
416 prompt "Transparent Hugepage Support sysfs defaults"
417 depends on TRANSPARENT_HUGEPAGE
418 default TRANSPARENT_HUGEPAGE_ALWAYS
420 Selects the sysfs defaults for Transparent Hugepage Support.
422 config TRANSPARENT_HUGEPAGE_ALWAYS
425 Enabling Transparent Hugepage always, can increase the
426 memory footprint of applications without a guaranteed
427 benefit but it will work automatically for all applications.
429 config TRANSPARENT_HUGEPAGE_MADVISE
432 Enabling Transparent Hugepage madvise, will only provide a
433 performance improvement benefit to the applications using
434 madvise(MADV_HUGEPAGE) but it won't risk to increase the
435 memory footprint of applications without a guaranteed
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"
451 Cleancache can be thought of as a page-granularity victim cache
452 for clean pages that the kernel's pageframe replacement algorithm
453 (PFRA) would like to keep around, but can't since there isn't enough
454 memory. So when the PFRA "evicts" a page, it first attempts to use
455 cleancache code to put the data contained in that page into
456 "transcendent memory", memory that is not directly accessible or
457 addressable by the kernel and is of unknown and possibly
458 time-varying size. And when a cleancache-enabled
459 filesystem wishes to access a page in a file on disk, it first
460 checks cleancache to see if it already contains it; if it does,
461 the page is copied into the kernel and a disk access is avoided.
462 When a transcendent memory driver is available (such as zcache or
463 Xen transcendent memory), a significant I/O reduction
464 may be achieved. When none is available, all cleancache calls
465 are reduced to a single pointer-compare-against-NULL resulting
466 in a negligible performance hit.
468 If unsure, say Y to enable cleancache
471 bool "Enable frontswap to cache swap pages if tmem is present"
475 Frontswap is so named because it can be thought of as the opposite
476 of a "backing" store for a swap device. The data is stored into
477 "transcendent memory", memory that is not directly accessible or
478 addressable by the kernel and is of unknown and possibly
479 time-varying size. When space in transcendent memory is available,
480 a significant swap I/O reduction may be achieved. When none is
481 available, all frontswap calls are reduced to a single pointer-
482 compare-against-NULL resulting in a negligible performance hit
483 and swap data is stored as normal on the matching swap device.
485 If unsure, say Y to enable frontswap.
488 bool "Contiguous Memory Allocator"
489 depends on HAVE_MEMBLOCK && MMU
491 select MEMORY_ISOLATION
493 This enables the Contiguous Memory Allocator which allows other
494 subsystems to allocate big physically-contiguous blocks of memory.
495 CMA reserves a region of memory and allows only movable pages to
496 be allocated from it. This way, the kernel can use the memory for
497 pagecache and when a subsystem requests for contiguous area, the
498 allocated pages are migrated away to serve the contiguous request.
503 bool "CMA debug messages (DEVELOPMENT)"
504 depends on DEBUG_KERNEL && CMA
506 Turns on debug messages in CMA. This produces KERN_DEBUG
507 messages for every CMA call as well as various messages while
508 processing calls such as dma_alloc_from_contiguous().
509 This option does not affect warning and error messages.
515 A special purpose allocator for storing compressed pages.
516 It is designed to store up to two compressed pages per physical
517 page. While this design limits storage density, it has simple and
518 deterministic reclaim properties that make it preferable to a higher
519 density approach when reclaim will be used.
522 bool "Compressed cache for swap pages (EXPERIMENTAL)"
523 depends on FRONTSWAP && CRYPTO=y
528 A lightweight compressed cache for swap pages. It takes
529 pages that are in the process of being swapped out and attempts to
530 compress them into a dynamically allocated RAM-based memory pool.
531 This can result in a significant I/O reduction on swap device and,
532 in the case where decompressing from RAM is faster that swap device
533 reads, can also improve workload performance.
535 This is marked experimental because it is a new feature (as of
536 v3.11) that interacts heavily with memory reclaim. While these
537 interactions don't cause any known issues on simple memory setups,
538 they have not be fully explored on the large set of potential
539 configurations and workloads that exist.
541 config MEM_SOFT_DIRTY
542 bool "Track memory changes"
543 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS
544 select PROC_PAGE_MONITOR
546 This option enables memory changes tracking by introducing a
547 soft-dirty bit on pte-s. This bit it set when someone writes
548 into a page just as regular dirty bit, but unlike the latter
549 it can be cleared by hands.
551 See Documentation/vm/soft-dirty.txt for more details.
554 tristate "Memory allocator for compressed pages"
558 zsmalloc is a slab-based memory allocator designed to store
559 compressed RAM pages. zsmalloc uses virtual memory mapping
560 in order to reduce fragmentation. However, this results in a
561 non-standard allocator interface where a handle, not a pointer, is
562 returned by an alloc(). This handle must be mapped in order to
563 access the allocated space.
565 config PGTABLE_MAPPING
566 bool "Use page table mapping to access object in zsmalloc"
569 By default, zsmalloc uses a copy-based object mapping method to
570 access allocations that span two pages. However, if a particular
571 architecture (ex, ARM) performs VM mapping faster than copying,
572 then you should select this. This causes zsmalloc to use page table
573 mapping rather than copying for object mapping.
575 You can check speed with zsmalloc benchmark:
576 https://github.com/spartacus06/zsmapbench
578 config GENERIC_EARLY_IOREMAP
581 config MAX_STACK_SIZE_MB
582 int "Maximum user stack size for 32-bit processes (MB)"
586 depends on STACK_GROWSUP && (!64BIT || COMPAT)
588 This is the maximum stack size in Megabytes in the VM layout of 32-bit
589 user processes when the stack grows upwards (currently only on parisc
590 and metag arch). The stack will be located at the highest memory
591 address minus the given value, unless the RLIMIT_STACK hard limit is
592 changed to a smaller value in which case that is used.
594 A sane initial value is 80 MB.