1ec8f24b7SThomas Gleixner# SPDX-License-Identifier: GPL-2.0-only 259e0b520SChristoph Hellwig 359e0b520SChristoph Hellwigmenu "Memory Management options" 459e0b520SChristoph Hellwig 5e1785e85SDave Hansenconfig SELECT_MEMORY_MODEL 6e1785e85SDave Hansen def_bool y 7a8826eebSKees Cook depends on ARCH_SELECT_MEMORY_MODEL 8e1785e85SDave Hansen 93a9da765SDave Hansenchoice 103a9da765SDave Hansen prompt "Memory model" 11e1785e85SDave Hansen depends on SELECT_MEMORY_MODEL 12e1785e85SDave Hansen default DISCONTIGMEM_MANUAL if ARCH_DISCONTIGMEM_DEFAULT 13d41dee36SAndy Whitcroft default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT 14e1785e85SDave Hansen default FLATMEM_MANUAL 15d66d109dSMike Rapoport help 16d66d109dSMike Rapoport This option allows you to change some of the ways that 17d66d109dSMike Rapoport Linux manages its memory internally. Most users will 18d66d109dSMike Rapoport only have one option here selected by the architecture 19d66d109dSMike Rapoport configuration. This is normal. 203a9da765SDave Hansen 21e1785e85SDave Hansenconfig FLATMEM_MANUAL 223a9da765SDave Hansen bool "Flat Memory" 23c898ec16SAnton Blanchard depends on !(ARCH_DISCONTIGMEM_ENABLE || ARCH_SPARSEMEM_ENABLE) || ARCH_FLATMEM_ENABLE 243a9da765SDave Hansen help 25d66d109dSMike Rapoport This option is best suited for non-NUMA systems with 26d66d109dSMike Rapoport flat address space. The FLATMEM is the most efficient 27d66d109dSMike Rapoport system in terms of performance and resource consumption 28d66d109dSMike Rapoport and it is the best option for smaller systems. 293a9da765SDave Hansen 30d66d109dSMike Rapoport For systems that have holes in their physical address 31d66d109dSMike Rapoport spaces and for features like NUMA and memory hotplug, 32dd33d29aSRandy Dunlap choose "Sparse Memory". 33d41dee36SAndy Whitcroft 34d41dee36SAndy Whitcroft If unsure, choose this option (Flat Memory) over any other. 353a9da765SDave Hansen 36e1785e85SDave Hansenconfig DISCONTIGMEM_MANUAL 37f3519f91SDave Hansen bool "Discontiguous Memory" 383a9da765SDave Hansen depends on ARCH_DISCONTIGMEM_ENABLE 393a9da765SDave Hansen help 40785dcd44SDave Hansen This option provides enhanced support for discontiguous 41785dcd44SDave Hansen memory systems, over FLATMEM. These systems have holes 42785dcd44SDave Hansen in their physical address spaces, and this option provides 43d66d109dSMike Rapoport more efficient handling of these holes. 44785dcd44SDave Hansen 45d66d109dSMike Rapoport Although "Discontiguous Memory" is still used by several 46d66d109dSMike Rapoport architectures, it is considered deprecated in favor of 47d66d109dSMike Rapoport "Sparse Memory". 48785dcd44SDave Hansen 49d66d109dSMike Rapoport If unsure, choose "Sparse Memory" over this option. 503a9da765SDave Hansen 51d41dee36SAndy Whitcroftconfig SPARSEMEM_MANUAL 52d41dee36SAndy Whitcroft bool "Sparse Memory" 53d41dee36SAndy Whitcroft depends on ARCH_SPARSEMEM_ENABLE 54d41dee36SAndy Whitcroft help 55d41dee36SAndy Whitcroft This will be the only option for some systems, including 56d66d109dSMike Rapoport memory hot-plug systems. This is normal. 57d41dee36SAndy Whitcroft 58d66d109dSMike Rapoport This option provides efficient support for systems with 59d66d109dSMike Rapoport holes is their physical address space and allows memory 60d66d109dSMike Rapoport hot-plug and hot-remove. 61d41dee36SAndy Whitcroft 62d66d109dSMike Rapoport If unsure, choose "Flat Memory" over this option. 63d41dee36SAndy Whitcroft 643a9da765SDave Hansenendchoice 653a9da765SDave Hansen 66e1785e85SDave Hansenconfig DISCONTIGMEM 67e1785e85SDave Hansen def_bool y 68e1785e85SDave Hansen depends on (!SELECT_MEMORY_MODEL && ARCH_DISCONTIGMEM_ENABLE) || DISCONTIGMEM_MANUAL 69e1785e85SDave Hansen 70d41dee36SAndy Whitcroftconfig SPARSEMEM 71d41dee36SAndy Whitcroft def_bool y 721a83e175SRussell King depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL 73d41dee36SAndy Whitcroft 74e1785e85SDave Hansenconfig FLATMEM 75e1785e85SDave Hansen def_bool y 76d41dee36SAndy Whitcroft depends on (!DISCONTIGMEM && !SPARSEMEM) || FLATMEM_MANUAL 77d41dee36SAndy Whitcroft 78d41dee36SAndy Whitcroftconfig FLAT_NODE_MEM_MAP 79d41dee36SAndy Whitcroft def_bool y 80d41dee36SAndy Whitcroft depends on !SPARSEMEM 81e1785e85SDave Hansen 8293b7504eSDave Hansen# 8393b7504eSDave Hansen# Both the NUMA code and DISCONTIGMEM use arrays of pg_data_t's 8493b7504eSDave Hansen# to represent different areas of memory. This variable allows 8593b7504eSDave Hansen# those dependencies to exist individually. 8693b7504eSDave Hansen# 8793b7504eSDave Hansenconfig NEED_MULTIPLE_NODES 8893b7504eSDave Hansen def_bool y 8993b7504eSDave Hansen depends on DISCONTIGMEM || NUMA 90af705362SAndy Whitcroft 91802f192eSBob Picco# 923e347261SBob Picco# SPARSEMEM_EXTREME (which is the default) does some bootmem 93c89ab04fSMike Rapoport# allocations when sparse_init() is called. If this cannot 943e347261SBob Picco# be done on your architecture, select this option. However, 953e347261SBob Picco# statically allocating the mem_section[] array can potentially 963e347261SBob Picco# consume vast quantities of .bss, so be careful. 973e347261SBob Picco# 983e347261SBob Picco# This option will also potentially produce smaller runtime code 993e347261SBob Picco# with gcc 3.4 and later. 1003e347261SBob Picco# 1013e347261SBob Piccoconfig SPARSEMEM_STATIC 1029ba16087SJan Beulich bool 1033e347261SBob Picco 1043e347261SBob Picco# 10544c09201SMatt LaPlante# Architecture platforms which require a two level mem_section in SPARSEMEM 106802f192eSBob Picco# must select this option. This is usually for architecture platforms with 107802f192eSBob Picco# an extremely sparse physical address space. 108802f192eSBob Picco# 1093e347261SBob Piccoconfig SPARSEMEM_EXTREME 1103e347261SBob Picco def_bool y 1113e347261SBob Picco depends on SPARSEMEM && !SPARSEMEM_STATIC 1124c21e2f2SHugh Dickins 11329c71111SAndy Whitcroftconfig SPARSEMEM_VMEMMAP_ENABLE 1149ba16087SJan Beulich bool 11529c71111SAndy Whitcroft 11629c71111SAndy Whitcroftconfig SPARSEMEM_VMEMMAP 117a5ee6daaSGeoff Levand bool "Sparse Memory virtual memmap" 118a5ee6daaSGeoff Levand depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE 119a5ee6daaSGeoff Levand default y 120a5ee6daaSGeoff Levand help 121a5ee6daaSGeoff Levand SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise 122a5ee6daaSGeoff Levand pfn_to_page and page_to_pfn operations. This is the most 123a5ee6daaSGeoff Levand efficient option when sufficient kernel resources are available. 12429c71111SAndy Whitcroft 12570210ed9SPhilipp Hachtmannconfig HAVE_MEMBLOCK_PHYS_MAP 1266341e62bSChristoph Jaeger bool 12770210ed9SPhilipp Hachtmann 12867a929e0SChristoph Hellwigconfig HAVE_FAST_GUP 129050a9adcSChristoph Hellwig depends on MMU 1306341e62bSChristoph Jaeger bool 1312667f50eSSteve Capper 13252219aeaSDavid Hildenbrand# Don't discard allocated memory used to track "memory" and "reserved" memblocks 13352219aeaSDavid Hildenbrand# after early boot, so it can still be used to test for validity of memory. 13452219aeaSDavid Hildenbrand# Also, memblocks are updated with memory hot(un)plug. 135350e88baSMike Rapoportconfig ARCH_KEEP_MEMBLOCK 1366341e62bSChristoph Jaeger bool 137c378ddd5STejun Heo 1381e5d8e1eSDan Williams# Keep arch NUMA mapping infrastructure post-init. 1391e5d8e1eSDan Williamsconfig NUMA_KEEP_MEMINFO 1401e5d8e1eSDan Williams bool 1411e5d8e1eSDan Williams 142ee6f509cSMinchan Kimconfig MEMORY_ISOLATION 1436341e62bSChristoph Jaeger bool 144ee6f509cSMinchan Kim 14546723bfaSYasuaki Ishimatsu# 14646723bfaSYasuaki Ishimatsu# Only be set on architectures that have completely implemented memory hotplug 14746723bfaSYasuaki Ishimatsu# feature. If you are not sure, don't touch it. 14846723bfaSYasuaki Ishimatsu# 14946723bfaSYasuaki Ishimatsuconfig HAVE_BOOTMEM_INFO_NODE 15046723bfaSYasuaki Ishimatsu def_bool n 15146723bfaSYasuaki Ishimatsu 1523947be19SDave Hansen# eventually, we can have this option just 'select SPARSEMEM' 1533947be19SDave Hansenconfig MEMORY_HOTPLUG 1543947be19SDave Hansen bool "Allow for memory hot-add" 155ec69acbbSKeith Mannthey depends on SPARSEMEM || X86_64_ACPI_NUMA 15640b31360SStephen Rothwell depends on ARCH_ENABLE_MEMORY_HOTPLUG 157b59d02edSMichal Hocko depends on 64BIT || BROKEN 1581e5d8e1eSDan Williams select NUMA_KEEP_MEMINFO if NUMA 1593947be19SDave Hansen 160ec69acbbSKeith Manntheyconfig MEMORY_HOTPLUG_SPARSE 161ec69acbbSKeith Mannthey def_bool y 162ec69acbbSKeith Mannthey depends on SPARSEMEM && MEMORY_HOTPLUG 163ec69acbbSKeith Mannthey 1648604d9e5SVitaly Kuznetsovconfig MEMORY_HOTPLUG_DEFAULT_ONLINE 1658604d9e5SVitaly Kuznetsov bool "Online the newly added memory blocks by default" 1668604d9e5SVitaly Kuznetsov depends on MEMORY_HOTPLUG 1678604d9e5SVitaly Kuznetsov help 1688604d9e5SVitaly Kuznetsov This option sets the default policy setting for memory hotplug 1698604d9e5SVitaly Kuznetsov onlining policy (/sys/devices/system/memory/auto_online_blocks) which 1708604d9e5SVitaly Kuznetsov determines what happens to newly added memory regions. Policy setting 1718604d9e5SVitaly Kuznetsov can always be changed at runtime. 172cb1aaebeSMauro Carvalho Chehab See Documentation/admin-guide/mm/memory-hotplug.rst for more information. 1738604d9e5SVitaly Kuznetsov 1748604d9e5SVitaly Kuznetsov Say Y here if you want all hot-plugged memory blocks to appear in 1758604d9e5SVitaly Kuznetsov 'online' state by default. 1768604d9e5SVitaly Kuznetsov Say N here if you want the default policy to keep all hot-plugged 1778604d9e5SVitaly Kuznetsov memory blocks in 'offline' state. 1788604d9e5SVitaly Kuznetsov 1790c0e6195SKAMEZAWA Hiroyukiconfig MEMORY_HOTREMOVE 1800c0e6195SKAMEZAWA Hiroyuki bool "Allow for memory hot remove" 18146723bfaSYasuaki Ishimatsu select MEMORY_ISOLATION 182f7e3334aSNathan Fontenot select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64) 1830c0e6195SKAMEZAWA Hiroyuki depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE 1840c0e6195SKAMEZAWA Hiroyuki depends on MIGRATION 1850c0e6195SKAMEZAWA Hiroyuki 1864c21e2f2SHugh Dickins# Heavily threaded applications may benefit from splitting the mm-wide 1874c21e2f2SHugh Dickins# page_table_lock, so that faults on different parts of the user address 1884c21e2f2SHugh Dickins# space can be handled with less contention: split it at this NR_CPUS. 1894c21e2f2SHugh Dickins# Default to 4 for wider testing, though 8 might be more appropriate. 1904c21e2f2SHugh Dickins# ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock. 1917b6ac9dfSHugh Dickins# PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes. 19260bccaa6SWill Deacon# SPARC32 allocates multiple pte tables within a single page, and therefore 19360bccaa6SWill Deacon# a per-page lock leads to problems when multiple tables need to be locked 19460bccaa6SWill Deacon# at the same time (e.g. copy_page_range()). 195a70caa8bSHugh Dickins# DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page. 1964c21e2f2SHugh Dickins# 1974c21e2f2SHugh Dickinsconfig SPLIT_PTLOCK_CPUS 1984c21e2f2SHugh Dickins int 1999164550eSKirill A. Shutemov default "999999" if !MMU 200a70caa8bSHugh Dickins default "999999" if ARM && !CPU_CACHE_VIPT 201a70caa8bSHugh Dickins default "999999" if PARISC && !PA20 20260bccaa6SWill Deacon default "999999" if SPARC32 2034c21e2f2SHugh Dickins default "4" 2047cbe34cfSChristoph Lameter 205e009bb30SKirill A. Shutemovconfig ARCH_ENABLE_SPLIT_PMD_PTLOCK 2066341e62bSChristoph Jaeger bool 207e009bb30SKirill A. Shutemov 2087cbe34cfSChristoph Lameter# 20909316c09SKonstantin Khlebnikov# support for memory balloon 21009316c09SKonstantin Khlebnikovconfig MEMORY_BALLOON 2116341e62bSChristoph Jaeger bool 21209316c09SKonstantin Khlebnikov 21309316c09SKonstantin Khlebnikov# 21418468d93SRafael Aquini# support for memory balloon compaction 21518468d93SRafael Aquiniconfig BALLOON_COMPACTION 21618468d93SRafael Aquini bool "Allow for balloon memory compaction/migration" 21718468d93SRafael Aquini def_bool y 21809316c09SKonstantin Khlebnikov depends on COMPACTION && MEMORY_BALLOON 21918468d93SRafael Aquini help 22018468d93SRafael Aquini Memory fragmentation introduced by ballooning might reduce 22118468d93SRafael Aquini significantly the number of 2MB contiguous memory blocks that can be 22218468d93SRafael Aquini used within a guest, thus imposing performance penalties associated 22318468d93SRafael Aquini with the reduced number of transparent huge pages that could be used 22418468d93SRafael Aquini by the guest workload. Allowing the compaction & migration for memory 22518468d93SRafael Aquini pages enlisted as being part of memory balloon devices avoids the 22618468d93SRafael Aquini scenario aforementioned and helps improving memory defragmentation. 22718468d93SRafael Aquini 22818468d93SRafael Aquini# 229e9e96b39SMel Gorman# support for memory compaction 230e9e96b39SMel Gormanconfig COMPACTION 231e9e96b39SMel Gorman bool "Allow for memory compaction" 23205106e6aSRik van Riel def_bool y 233e9e96b39SMel Gorman select MIGRATION 23433a93877SAndrea Arcangeli depends on MMU 235e9e96b39SMel Gorman help 236b32eaf71SMichal Hocko Compaction is the only memory management component to form 237b32eaf71SMichal Hocko high order (larger physically contiguous) memory blocks 238b32eaf71SMichal Hocko reliably. The page allocator relies on compaction heavily and 239b32eaf71SMichal Hocko the lack of the feature can lead to unexpected OOM killer 240b32eaf71SMichal Hocko invocations for high order memory requests. You shouldn't 241b32eaf71SMichal Hocko disable this option unless there really is a strong reason for 242b32eaf71SMichal Hocko it and then we would be really interested to hear about that at 243b32eaf71SMichal Hocko linux-mm@kvack.org. 244e9e96b39SMel Gorman 245e9e96b39SMel Gorman# 24636e66c55SAlexander Duyck# support for free page reporting 24736e66c55SAlexander Duyckconfig PAGE_REPORTING 24836e66c55SAlexander Duyck bool "Free page reporting" 24936e66c55SAlexander Duyck def_bool n 25036e66c55SAlexander Duyck help 25136e66c55SAlexander Duyck Free page reporting allows for the incremental acquisition of 25236e66c55SAlexander Duyck free pages from the buddy allocator for the purpose of reporting 25336e66c55SAlexander Duyck those pages to another entity, such as a hypervisor, so that the 25436e66c55SAlexander Duyck memory can be freed within the host for other uses. 25536e66c55SAlexander Duyck 25636e66c55SAlexander Duyck# 2577cbe34cfSChristoph Lameter# support for page migration 2587cbe34cfSChristoph Lameter# 2597cbe34cfSChristoph Lameterconfig MIGRATION 260b20a3503SChristoph Lameter bool "Page migration" 2616c5240aeSChristoph Lameter def_bool y 262de32a817SChen Gang depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU 263b20a3503SChristoph Lameter help 264b20a3503SChristoph Lameter Allows the migration of the physical location of pages of processes 265e9e96b39SMel Gorman while the virtual addresses are not changed. This is useful in 266e9e96b39SMel Gorman two situations. The first is on NUMA systems to put pages nearer 267e9e96b39SMel Gorman to the processors accessing. The second is when allocating huge 268e9e96b39SMel Gorman pages as migration can relocate pages to satisfy a huge page 269e9e96b39SMel Gorman allocation instead of reclaiming. 2706550e07fSGreg Kroah-Hartman 271c177c81eSNaoya Horiguchiconfig ARCH_ENABLE_HUGEPAGE_MIGRATION 2726341e62bSChristoph Jaeger bool 273c177c81eSNaoya Horiguchi 2749c670ea3SNaoya Horiguchiconfig ARCH_ENABLE_THP_MIGRATION 2759c670ea3SNaoya Horiguchi bool 2769c670ea3SNaoya Horiguchi 2778df995f6SAlexandre Ghiticonfig CONTIG_ALLOC 2788df995f6SAlexandre Ghiti def_bool (MEMORY_ISOLATION && COMPACTION) || CMA 2798df995f6SAlexandre Ghiti 280600715dcSJeremy Fitzhardingeconfig PHYS_ADDR_T_64BIT 281d4a451d5SChristoph Hellwig def_bool 64BIT 282600715dcSJeremy Fitzhardinge 2832a7326b5SChristoph Lameterconfig BOUNCE 2849ca24e2eSVinayak Menon bool "Enable bounce buffers" 2859ca24e2eSVinayak Menon default y 2862a7326b5SChristoph Lameter depends on BLOCK && MMU && (ZONE_DMA || HIGHMEM) 2879ca24e2eSVinayak Menon help 2889ca24e2eSVinayak Menon Enable bounce buffers for devices that cannot access 2899ca24e2eSVinayak Menon the full range of memory available to the CPU. Enabled 2909ca24e2eSVinayak Menon by default when ZONE_DMA or HIGHMEM is selected, but you 2919ca24e2eSVinayak Menon may say n to override this. 2922a7326b5SChristoph Lameter 293f057eac0SStephen Rothwellconfig VIRT_TO_BUS 2944febd95aSStephen Rothwell bool 2954febd95aSStephen Rothwell help 2964febd95aSStephen Rothwell An architecture should select this if it implements the 2974febd95aSStephen Rothwell deprecated interface virt_to_bus(). All new architectures 2984febd95aSStephen Rothwell should probably not select this. 2994febd95aSStephen Rothwell 300cddb8a5cSAndrea Arcangeli 301cddb8a5cSAndrea Arcangeliconfig MMU_NOTIFIER 302cddb8a5cSAndrea Arcangeli bool 30383fe27eaSPranith Kumar select SRCU 30499cb252fSJason Gunthorpe select INTERVAL_TREE 305fc4d5c29SDavid Howells 306f8af4da3SHugh Dickinsconfig KSM 307f8af4da3SHugh Dickins bool "Enable KSM for page merging" 308f8af4da3SHugh Dickins depends on MMU 30959e1a2f4STimofey Titovets select XXHASH 310f8af4da3SHugh Dickins help 311f8af4da3SHugh Dickins Enable Kernel Samepage Merging: KSM periodically scans those areas 312f8af4da3SHugh Dickins of an application's address space that an app has advised may be 313f8af4da3SHugh Dickins mergeable. When it finds pages of identical content, it replaces 314d0f209f6SHugh Dickins the many instances by a single page with that content, so 315f8af4da3SHugh Dickins saving memory until one or another app needs to modify the content. 316f8af4da3SHugh Dickins Recommended for use with KVM, or with other duplicative applications. 317ad56b738SMike Rapoport See Documentation/vm/ksm.rst for more information: KSM is inactive 318c73602adSHugh Dickins until a program has madvised that an area is MADV_MERGEABLE, and 319c73602adSHugh Dickins root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). 320f8af4da3SHugh Dickins 321e0a94c2aSChristoph Lameterconfig DEFAULT_MMAP_MIN_ADDR 322e0a94c2aSChristoph Lameter int "Low address space to protect from user allocation" 3236e141546SDavid Howells depends on MMU 324e0a94c2aSChristoph Lameter default 4096 325e0a94c2aSChristoph Lameter help 326e0a94c2aSChristoph Lameter This is the portion of low virtual memory which should be protected 327e0a94c2aSChristoph Lameter from userspace allocation. Keeping a user from writing to low pages 328e0a94c2aSChristoph Lameter can help reduce the impact of kernel NULL pointer bugs. 329e0a94c2aSChristoph Lameter 330e0a94c2aSChristoph Lameter For most ia64, ppc64 and x86 users with lots of address space 331e0a94c2aSChristoph Lameter a value of 65536 is reasonable and should cause no problems. 332e0a94c2aSChristoph Lameter On arm and other archs it should not be higher than 32768. 333788084abSEric Paris Programs which use vm86 functionality or have some need to map 334788084abSEric Paris this low address space will need CAP_SYS_RAWIO or disable this 335788084abSEric Paris protection by setting the value to 0. 336e0a94c2aSChristoph Lameter 337e0a94c2aSChristoph Lameter This value can be changed after boot using the 338e0a94c2aSChristoph Lameter /proc/sys/vm/mmap_min_addr tunable. 339e0a94c2aSChristoph Lameter 340d949f36fSLinus Torvaldsconfig ARCH_SUPPORTS_MEMORY_FAILURE 341d949f36fSLinus Torvalds bool 342e0a94c2aSChristoph Lameter 3436a46079cSAndi Kleenconfig MEMORY_FAILURE 3446a46079cSAndi Kleen depends on MMU 345d949f36fSLinus Torvalds depends on ARCH_SUPPORTS_MEMORY_FAILURE 3466a46079cSAndi Kleen bool "Enable recovery from hardware memory errors" 347ee6f509cSMinchan Kim select MEMORY_ISOLATION 34897f0b134SXie XiuQi select RAS 3496a46079cSAndi Kleen help 3506a46079cSAndi Kleen Enables code to recover from some memory failures on systems 3516a46079cSAndi Kleen with MCA recovery. This allows a system to continue running 3526a46079cSAndi Kleen even when some of its memory has uncorrected errors. This requires 3536a46079cSAndi Kleen special hardware support and typically ECC memory. 3546a46079cSAndi Kleen 355cae681fcSAndi Kleenconfig HWPOISON_INJECT 356413f9efbSAndi Kleen tristate "HWPoison pages injector" 35727df5068SAndi Kleen depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS 358478c5ffcSWu Fengguang select PROC_PAGE_MONITOR 359cae681fcSAndi Kleen 360fc4d5c29SDavid Howellsconfig NOMMU_INITIAL_TRIM_EXCESS 361fc4d5c29SDavid Howells int "Turn on mmap() excess space trimming before booting" 362fc4d5c29SDavid Howells depends on !MMU 363fc4d5c29SDavid Howells default 1 364fc4d5c29SDavid Howells help 365fc4d5c29SDavid Howells The NOMMU mmap() frequently needs to allocate large contiguous chunks 366fc4d5c29SDavid Howells of memory on which to store mappings, but it can only ask the system 367fc4d5c29SDavid Howells allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently 368fc4d5c29SDavid Howells more than it requires. To deal with this, mmap() is able to trim off 369fc4d5c29SDavid Howells the excess and return it to the allocator. 370fc4d5c29SDavid Howells 371fc4d5c29SDavid Howells If trimming is enabled, the excess is trimmed off and returned to the 372fc4d5c29SDavid Howells system allocator, which can cause extra fragmentation, particularly 373fc4d5c29SDavid Howells if there are a lot of transient processes. 374fc4d5c29SDavid Howells 375fc4d5c29SDavid Howells If trimming is disabled, the excess is kept, but not used, which for 376fc4d5c29SDavid Howells long-term mappings means that the space is wasted. 377fc4d5c29SDavid Howells 378fc4d5c29SDavid Howells Trimming can be dynamically controlled through a sysctl option 379fc4d5c29SDavid Howells (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of 380fc4d5c29SDavid Howells excess pages there must be before trimming should occur, or zero if 381fc4d5c29SDavid Howells no trimming is to occur. 382fc4d5c29SDavid Howells 383fc4d5c29SDavid Howells This option specifies the initial value of this option. The default 384fc4d5c29SDavid Howells of 1 says that all excess pages should be trimmed. 385fc4d5c29SDavid Howells 386*dd19d293SStephen Kitt See Documentation/admin-guide/mm/nommu-mmap.rst for more information. 387bbddff05STejun Heo 3884c76d9d1SAndrea Arcangeliconfig TRANSPARENT_HUGEPAGE 38913ece886SAndrea Arcangeli bool "Transparent Hugepage Support" 39015626062SGerald Schaefer depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE 3915d689240SAndrea Arcangeli select COMPACTION 3923a08cd52SMatthew Wilcox select XARRAY_MULTI 3934c76d9d1SAndrea Arcangeli help 3944c76d9d1SAndrea Arcangeli Transparent Hugepages allows the kernel to use huge pages and 3954c76d9d1SAndrea Arcangeli huge tlb transparently to the applications whenever possible. 3964c76d9d1SAndrea Arcangeli This feature can improve computing performance to certain 3974c76d9d1SAndrea Arcangeli applications by speeding up page faults during memory 3984c76d9d1SAndrea Arcangeli allocation, by reducing the number of tlb misses and by speeding 3994c76d9d1SAndrea Arcangeli up the pagetable walking. 4004c76d9d1SAndrea Arcangeli 4014c76d9d1SAndrea Arcangeli If memory constrained on embedded, you may want to say N. 4024c76d9d1SAndrea Arcangeli 40313ece886SAndrea Arcangelichoice 40413ece886SAndrea Arcangeli prompt "Transparent Hugepage Support sysfs defaults" 40513ece886SAndrea Arcangeli depends on TRANSPARENT_HUGEPAGE 40613ece886SAndrea Arcangeli default TRANSPARENT_HUGEPAGE_ALWAYS 40713ece886SAndrea Arcangeli help 40813ece886SAndrea Arcangeli Selects the sysfs defaults for Transparent Hugepage Support. 40913ece886SAndrea Arcangeli 41013ece886SAndrea Arcangeli config TRANSPARENT_HUGEPAGE_ALWAYS 41113ece886SAndrea Arcangeli bool "always" 41213ece886SAndrea Arcangeli help 41313ece886SAndrea Arcangeli Enabling Transparent Hugepage always, can increase the 41413ece886SAndrea Arcangeli memory footprint of applications without a guaranteed 41513ece886SAndrea Arcangeli benefit but it will work automatically for all applications. 41613ece886SAndrea Arcangeli 41713ece886SAndrea Arcangeli config TRANSPARENT_HUGEPAGE_MADVISE 41813ece886SAndrea Arcangeli bool "madvise" 41913ece886SAndrea Arcangeli help 42013ece886SAndrea Arcangeli Enabling Transparent Hugepage madvise, will only provide a 42113ece886SAndrea Arcangeli performance improvement benefit to the applications using 42213ece886SAndrea Arcangeli madvise(MADV_HUGEPAGE) but it won't risk to increase the 42313ece886SAndrea Arcangeli memory footprint of applications without a guaranteed 42413ece886SAndrea Arcangeli benefit. 42513ece886SAndrea Arcangeliendchoice 42613ece886SAndrea Arcangeli 42738d8b4e6SHuang Yingconfig ARCH_WANTS_THP_SWAP 42838d8b4e6SHuang Ying def_bool n 42938d8b4e6SHuang Ying 43038d8b4e6SHuang Yingconfig THP_SWAP 43138d8b4e6SHuang Ying def_bool y 43214fef284SHuang Ying depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP 43338d8b4e6SHuang Ying help 43438d8b4e6SHuang Ying Swap transparent huge pages in one piece, without splitting. 43514fef284SHuang Ying XXX: For now, swap cluster backing transparent huge page 43614fef284SHuang Ying will be split after swapout. 43738d8b4e6SHuang Ying 43838d8b4e6SHuang Ying For selection by architectures with reasonable THP sizes. 43938d8b4e6SHuang Ying 440e496cf3dSKirill A. Shutemov# 441bbddff05STejun Heo# UP and nommu archs use km based percpu allocator 442bbddff05STejun Heo# 443bbddff05STejun Heoconfig NEED_PER_CPU_KM 444bbddff05STejun Heo depends on !SMP 445bbddff05STejun Heo bool 446bbddff05STejun Heo default y 447077b1f83SDan Magenheimer 448077b1f83SDan Magenheimerconfig CLEANCACHE 449077b1f83SDan Magenheimer bool "Enable cleancache driver to cache clean pages if tmem is present" 450077b1f83SDan Magenheimer help 451077b1f83SDan Magenheimer Cleancache can be thought of as a page-granularity victim cache 452077b1f83SDan Magenheimer for clean pages that the kernel's pageframe replacement algorithm 453077b1f83SDan Magenheimer (PFRA) would like to keep around, but can't since there isn't enough 454077b1f83SDan Magenheimer memory. So when the PFRA "evicts" a page, it first attempts to use 455140a1ef2SMichael Witten cleancache code to put the data contained in that page into 456077b1f83SDan Magenheimer "transcendent memory", memory that is not directly accessible or 457077b1f83SDan Magenheimer addressable by the kernel and is of unknown and possibly 458077b1f83SDan Magenheimer time-varying size. And when a cleancache-enabled 459077b1f83SDan Magenheimer filesystem wishes to access a page in a file on disk, it first 460077b1f83SDan Magenheimer checks cleancache to see if it already contains it; if it does, 461077b1f83SDan Magenheimer the page is copied into the kernel and a disk access is avoided. 462077b1f83SDan Magenheimer When a transcendent memory driver is available (such as zcache or 463077b1f83SDan Magenheimer Xen transcendent memory), a significant I/O reduction 464077b1f83SDan Magenheimer may be achieved. When none is available, all cleancache calls 465077b1f83SDan Magenheimer are reduced to a single pointer-compare-against-NULL resulting 466077b1f83SDan Magenheimer in a negligible performance hit. 467077b1f83SDan Magenheimer 468077b1f83SDan Magenheimer If unsure, say Y to enable cleancache 46927c6aec2SDan Magenheimer 47027c6aec2SDan Magenheimerconfig FRONTSWAP 47127c6aec2SDan Magenheimer bool "Enable frontswap to cache swap pages if tmem is present" 47227c6aec2SDan Magenheimer depends on SWAP 47327c6aec2SDan Magenheimer help 47427c6aec2SDan Magenheimer Frontswap is so named because it can be thought of as the opposite 47527c6aec2SDan Magenheimer of a "backing" store for a swap device. The data is stored into 47627c6aec2SDan Magenheimer "transcendent memory", memory that is not directly accessible or 47727c6aec2SDan Magenheimer addressable by the kernel and is of unknown and possibly 47827c6aec2SDan Magenheimer time-varying size. When space in transcendent memory is available, 47927c6aec2SDan Magenheimer a significant swap I/O reduction may be achieved. When none is 48027c6aec2SDan Magenheimer available, all frontswap calls are reduced to a single pointer- 48127c6aec2SDan Magenheimer compare-against-NULL resulting in a negligible performance hit 48227c6aec2SDan Magenheimer and swap data is stored as normal on the matching swap device. 48327c6aec2SDan Magenheimer 48427c6aec2SDan Magenheimer If unsure, say Y to enable frontswap. 485f825c736SAneesh Kumar K.V 486f825c736SAneesh Kumar K.Vconfig CMA 487f825c736SAneesh Kumar K.V bool "Contiguous Memory Allocator" 488aca52c39SMike Rapoport depends on MMU 489f825c736SAneesh Kumar K.V select MIGRATION 490f825c736SAneesh Kumar K.V select MEMORY_ISOLATION 491f825c736SAneesh Kumar K.V help 492f825c736SAneesh Kumar K.V This enables the Contiguous Memory Allocator which allows other 493f825c736SAneesh Kumar K.V subsystems to allocate big physically-contiguous blocks of memory. 494f825c736SAneesh Kumar K.V CMA reserves a region of memory and allows only movable pages to 495f825c736SAneesh Kumar K.V be allocated from it. This way, the kernel can use the memory for 496f825c736SAneesh Kumar K.V pagecache and when a subsystem requests for contiguous area, the 497f825c736SAneesh Kumar K.V allocated pages are migrated away to serve the contiguous request. 498f825c736SAneesh Kumar K.V 499f825c736SAneesh Kumar K.V If unsure, say "n". 500f825c736SAneesh Kumar K.V 501f825c736SAneesh Kumar K.Vconfig CMA_DEBUG 502f825c736SAneesh Kumar K.V bool "CMA debug messages (DEVELOPMENT)" 503f825c736SAneesh Kumar K.V depends on DEBUG_KERNEL && CMA 504f825c736SAneesh Kumar K.V help 505f825c736SAneesh Kumar K.V Turns on debug messages in CMA. This produces KERN_DEBUG 506f825c736SAneesh Kumar K.V messages for every CMA call as well as various messages while 507f825c736SAneesh Kumar K.V processing calls such as dma_alloc_from_contiguous(). 508f825c736SAneesh Kumar K.V This option does not affect warning and error messages. 509bf550fc9SAlexander Graf 51028b24c1fSSasha Levinconfig CMA_DEBUGFS 51128b24c1fSSasha Levin bool "CMA debugfs interface" 51228b24c1fSSasha Levin depends on CMA && DEBUG_FS 51328b24c1fSSasha Levin help 51428b24c1fSSasha Levin Turns on the DebugFS interface for CMA. 51528b24c1fSSasha Levin 516a254129eSJoonsoo Kimconfig CMA_AREAS 517a254129eSJoonsoo Kim int "Maximum count of the CMA areas" 518a254129eSJoonsoo Kim depends on CMA 519a254129eSJoonsoo Kim default 7 520a254129eSJoonsoo Kim help 521a254129eSJoonsoo Kim CMA allows to create CMA areas for particular purpose, mainly, 522a254129eSJoonsoo Kim used as device private area. This parameter sets the maximum 523a254129eSJoonsoo Kim number of CMA area in the system. 524a254129eSJoonsoo Kim 525a254129eSJoonsoo Kim If unsure, leave the default value "7". 526a254129eSJoonsoo Kim 527af8d417aSDan Streetmanconfig MEM_SOFT_DIRTY 528af8d417aSDan Streetman bool "Track memory changes" 529af8d417aSDan Streetman depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS 530af8d417aSDan Streetman select PROC_PAGE_MONITOR 5314e2e2770SSeth Jennings help 532af8d417aSDan Streetman This option enables memory changes tracking by introducing a 533af8d417aSDan Streetman soft-dirty bit on pte-s. This bit it set when someone writes 534af8d417aSDan Streetman into a page just as regular dirty bit, but unlike the latter 535af8d417aSDan Streetman it can be cleared by hands. 536af8d417aSDan Streetman 5371ad1335dSMike Rapoport See Documentation/admin-guide/mm/soft-dirty.rst for more details. 5384e2e2770SSeth Jennings 5392b281117SSeth Jenningsconfig ZSWAP 5402b281117SSeth Jennings bool "Compressed cache for swap pages (EXPERIMENTAL)" 5412b281117SSeth Jennings depends on FRONTSWAP && CRYPTO=y 54212d79d64SDan Streetman select ZPOOL 5432b281117SSeth Jennings help 5442b281117SSeth Jennings A lightweight compressed cache for swap pages. It takes 5452b281117SSeth Jennings pages that are in the process of being swapped out and attempts to 5462b281117SSeth Jennings compress them into a dynamically allocated RAM-based memory pool. 5472b281117SSeth Jennings This can result in a significant I/O reduction on swap device and, 5482b281117SSeth Jennings in the case where decompressing from RAM is faster that swap device 5492b281117SSeth Jennings reads, can also improve workload performance. 5502b281117SSeth Jennings 5512b281117SSeth Jennings This is marked experimental because it is a new feature (as of 5522b281117SSeth Jennings v3.11) that interacts heavily with memory reclaim. While these 5532b281117SSeth Jennings interactions don't cause any known issues on simple memory setups, 5542b281117SSeth Jennings they have not be fully explored on the large set of potential 5552b281117SSeth Jennings configurations and workloads that exist. 5562b281117SSeth Jennings 557bb8b93b5SMaciej S. Szmigierochoice 558bb8b93b5SMaciej S. Szmigiero prompt "Compressed cache for swap pages default compressor" 559bb8b93b5SMaciej S. Szmigiero depends on ZSWAP 560bb8b93b5SMaciej S. Szmigiero default ZSWAP_COMPRESSOR_DEFAULT_LZO 561bb8b93b5SMaciej S. Szmigiero help 562bb8b93b5SMaciej S. Szmigiero Selects the default compression algorithm for the compressed cache 563bb8b93b5SMaciej S. Szmigiero for swap pages. 564bb8b93b5SMaciej S. Szmigiero 565bb8b93b5SMaciej S. Szmigiero For an overview what kind of performance can be expected from 566bb8b93b5SMaciej S. Szmigiero a particular compression algorithm please refer to the benchmarks 567bb8b93b5SMaciej S. Szmigiero available at the following LWN page: 568bb8b93b5SMaciej S. Szmigiero https://lwn.net/Articles/751795/ 569bb8b93b5SMaciej S. Szmigiero 570bb8b93b5SMaciej S. Szmigiero If in doubt, select 'LZO'. 571bb8b93b5SMaciej S. Szmigiero 572bb8b93b5SMaciej S. Szmigiero The selection made here can be overridden by using the kernel 573bb8b93b5SMaciej S. Szmigiero command line 'zswap.compressor=' option. 574bb8b93b5SMaciej S. Szmigiero 575bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_COMPRESSOR_DEFAULT_DEFLATE 576bb8b93b5SMaciej S. Szmigiero bool "Deflate" 577bb8b93b5SMaciej S. Szmigiero select CRYPTO_DEFLATE 578bb8b93b5SMaciej S. Szmigiero help 579bb8b93b5SMaciej S. Szmigiero Use the Deflate algorithm as the default compression algorithm. 580bb8b93b5SMaciej S. Szmigiero 581bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_COMPRESSOR_DEFAULT_LZO 582bb8b93b5SMaciej S. Szmigiero bool "LZO" 583bb8b93b5SMaciej S. Szmigiero select CRYPTO_LZO 584bb8b93b5SMaciej S. Szmigiero help 585bb8b93b5SMaciej S. Szmigiero Use the LZO algorithm as the default compression algorithm. 586bb8b93b5SMaciej S. Szmigiero 587bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_COMPRESSOR_DEFAULT_842 588bb8b93b5SMaciej S. Szmigiero bool "842" 589bb8b93b5SMaciej S. Szmigiero select CRYPTO_842 590bb8b93b5SMaciej S. Szmigiero help 591bb8b93b5SMaciej S. Szmigiero Use the 842 algorithm as the default compression algorithm. 592bb8b93b5SMaciej S. Szmigiero 593bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_COMPRESSOR_DEFAULT_LZ4 594bb8b93b5SMaciej S. Szmigiero bool "LZ4" 595bb8b93b5SMaciej S. Szmigiero select CRYPTO_LZ4 596bb8b93b5SMaciej S. Szmigiero help 597bb8b93b5SMaciej S. Szmigiero Use the LZ4 algorithm as the default compression algorithm. 598bb8b93b5SMaciej S. Szmigiero 599bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_COMPRESSOR_DEFAULT_LZ4HC 600bb8b93b5SMaciej S. Szmigiero bool "LZ4HC" 601bb8b93b5SMaciej S. Szmigiero select CRYPTO_LZ4HC 602bb8b93b5SMaciej S. Szmigiero help 603bb8b93b5SMaciej S. Szmigiero Use the LZ4HC algorithm as the default compression algorithm. 604bb8b93b5SMaciej S. Szmigiero 605bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_COMPRESSOR_DEFAULT_ZSTD 606bb8b93b5SMaciej S. Szmigiero bool "zstd" 607bb8b93b5SMaciej S. Szmigiero select CRYPTO_ZSTD 608bb8b93b5SMaciej S. Szmigiero help 609bb8b93b5SMaciej S. Szmigiero Use the zstd algorithm as the default compression algorithm. 610bb8b93b5SMaciej S. Szmigieroendchoice 611bb8b93b5SMaciej S. Szmigiero 612bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_COMPRESSOR_DEFAULT 613bb8b93b5SMaciej S. Szmigiero string 614bb8b93b5SMaciej S. Szmigiero depends on ZSWAP 615bb8b93b5SMaciej S. Szmigiero default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE 616bb8b93b5SMaciej S. Szmigiero default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO 617bb8b93b5SMaciej S. Szmigiero default "842" if ZSWAP_COMPRESSOR_DEFAULT_842 618bb8b93b5SMaciej S. Szmigiero default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4 619bb8b93b5SMaciej S. Szmigiero default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC 620bb8b93b5SMaciej S. Szmigiero default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD 621bb8b93b5SMaciej S. Szmigiero default "" 622bb8b93b5SMaciej S. Szmigiero 623bb8b93b5SMaciej S. Szmigierochoice 624bb8b93b5SMaciej S. Szmigiero prompt "Compressed cache for swap pages default allocator" 625bb8b93b5SMaciej S. Szmigiero depends on ZSWAP 626bb8b93b5SMaciej S. Szmigiero default ZSWAP_ZPOOL_DEFAULT_ZBUD 627bb8b93b5SMaciej S. Szmigiero help 628bb8b93b5SMaciej S. Szmigiero Selects the default allocator for the compressed cache for 629bb8b93b5SMaciej S. Szmigiero swap pages. 630bb8b93b5SMaciej S. Szmigiero The default is 'zbud' for compatibility, however please do 631bb8b93b5SMaciej S. Szmigiero read the description of each of the allocators below before 632bb8b93b5SMaciej S. Szmigiero making a right choice. 633bb8b93b5SMaciej S. Szmigiero 634bb8b93b5SMaciej S. Szmigiero The selection made here can be overridden by using the kernel 635bb8b93b5SMaciej S. Szmigiero command line 'zswap.zpool=' option. 636bb8b93b5SMaciej S. Szmigiero 637bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_ZPOOL_DEFAULT_ZBUD 638bb8b93b5SMaciej S. Szmigiero bool "zbud" 639bb8b93b5SMaciej S. Szmigiero select ZBUD 640bb8b93b5SMaciej S. Szmigiero help 641bb8b93b5SMaciej S. Szmigiero Use the zbud allocator as the default allocator. 642bb8b93b5SMaciej S. Szmigiero 643bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_ZPOOL_DEFAULT_Z3FOLD 644bb8b93b5SMaciej S. Szmigiero bool "z3fold" 645bb8b93b5SMaciej S. Szmigiero select Z3FOLD 646bb8b93b5SMaciej S. Szmigiero help 647bb8b93b5SMaciej S. Szmigiero Use the z3fold allocator as the default allocator. 648bb8b93b5SMaciej S. Szmigiero 649bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_ZPOOL_DEFAULT_ZSMALLOC 650bb8b93b5SMaciej S. Szmigiero bool "zsmalloc" 651bb8b93b5SMaciej S. Szmigiero select ZSMALLOC 652bb8b93b5SMaciej S. Szmigiero help 653bb8b93b5SMaciej S. Szmigiero Use the zsmalloc allocator as the default allocator. 654bb8b93b5SMaciej S. Szmigieroendchoice 655bb8b93b5SMaciej S. Szmigiero 656bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_ZPOOL_DEFAULT 657bb8b93b5SMaciej S. Szmigiero string 658bb8b93b5SMaciej S. Szmigiero depends on ZSWAP 659bb8b93b5SMaciej S. Szmigiero default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD 660bb8b93b5SMaciej S. Szmigiero default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD 661bb8b93b5SMaciej S. Szmigiero default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC 662bb8b93b5SMaciej S. Szmigiero default "" 663bb8b93b5SMaciej S. Szmigiero 664bb8b93b5SMaciej S. Szmigieroconfig ZSWAP_DEFAULT_ON 665bb8b93b5SMaciej S. Szmigiero bool "Enable the compressed cache for swap pages by default" 666bb8b93b5SMaciej S. Szmigiero depends on ZSWAP 667bb8b93b5SMaciej S. Szmigiero help 668bb8b93b5SMaciej S. Szmigiero If selected, the compressed cache for swap pages will be enabled 669bb8b93b5SMaciej S. Szmigiero at boot, otherwise it will be disabled. 670bb8b93b5SMaciej S. Szmigiero 671bb8b93b5SMaciej S. Szmigiero The selection made here can be overridden by using the kernel 672bb8b93b5SMaciej S. Szmigiero command line 'zswap.enabled=' option. 673bb8b93b5SMaciej S. Szmigiero 674af8d417aSDan Streetmanconfig ZPOOL 675af8d417aSDan Streetman tristate "Common API for compressed memory storage" 6760f8975ecSPavel Emelyanov help 677af8d417aSDan Streetman Compressed memory storage API. This allows using either zbud or 678af8d417aSDan Streetman zsmalloc. 6790f8975ecSPavel Emelyanov 680af8d417aSDan Streetmanconfig ZBUD 6819a001fc1SVitaly Wool tristate "Low (Up to 2x) density storage for compressed pages" 682af8d417aSDan Streetman help 683af8d417aSDan Streetman A special purpose allocator for storing compressed pages. 684af8d417aSDan Streetman It is designed to store up to two compressed pages per physical 685af8d417aSDan Streetman page. While this design limits storage density, it has simple and 686af8d417aSDan Streetman deterministic reclaim properties that make it preferable to a higher 687af8d417aSDan Streetman density approach when reclaim will be used. 688bcf1647dSMinchan Kim 6899a001fc1SVitaly Woolconfig Z3FOLD 6909a001fc1SVitaly Wool tristate "Up to 3x density storage for compressed pages" 6919a001fc1SVitaly Wool depends on ZPOOL 6929a001fc1SVitaly Wool help 6939a001fc1SVitaly Wool A special purpose allocator for storing compressed pages. 6949a001fc1SVitaly Wool It is designed to store up to three compressed pages per physical 6959a001fc1SVitaly Wool page. It is a ZBUD derivative so the simplicity and determinism are 6969a001fc1SVitaly Wool still there. 6979a001fc1SVitaly Wool 698bcf1647dSMinchan Kimconfig ZSMALLOC 699d867f203SMinchan Kim tristate "Memory allocator for compressed pages" 700bcf1647dSMinchan Kim depends on MMU 701bcf1647dSMinchan Kim help 702bcf1647dSMinchan Kim zsmalloc is a slab-based memory allocator designed to store 703bcf1647dSMinchan Kim compressed RAM pages. zsmalloc uses virtual memory mapping 704bcf1647dSMinchan Kim in order to reduce fragmentation. However, this results in a 705bcf1647dSMinchan Kim non-standard allocator interface where a handle, not a pointer, is 706bcf1647dSMinchan Kim returned by an alloc(). This handle must be mapped in order to 707bcf1647dSMinchan Kim access the allocated space. 708bcf1647dSMinchan Kim 7098b136018SChristoph Hellwigconfig ZSMALLOC_PGTABLE_MAPPING 710bcf1647dSMinchan Kim bool "Use page table mapping to access object in zsmalloc" 711b607e6d1SChristoph Hellwig depends on ZSMALLOC=y 712bcf1647dSMinchan Kim help 713bcf1647dSMinchan Kim By default, zsmalloc uses a copy-based object mapping method to 714bcf1647dSMinchan Kim access allocations that span two pages. However, if a particular 715bcf1647dSMinchan Kim architecture (ex, ARM) performs VM mapping faster than copying, 716bcf1647dSMinchan Kim then you should select this. This causes zsmalloc to use page table 717bcf1647dSMinchan Kim mapping rather than copying for object mapping. 718bcf1647dSMinchan Kim 7192216ee85SBen Hutchings You can check speed with zsmalloc benchmark: 7202216ee85SBen Hutchings https://github.com/spartacus06/zsmapbench 7219e5c33d7SMark Salter 7220f050d99SGanesh Mahendranconfig ZSMALLOC_STAT 7230f050d99SGanesh Mahendran bool "Export zsmalloc statistics" 7240f050d99SGanesh Mahendran depends on ZSMALLOC 7250f050d99SGanesh Mahendran select DEBUG_FS 7260f050d99SGanesh Mahendran help 7270f050d99SGanesh Mahendran This option enables code in the zsmalloc to collect various 7280f050d99SGanesh Mahendran statistics about whats happening in zsmalloc and exports that 7290f050d99SGanesh Mahendran information to userspace via debugfs. 7300f050d99SGanesh Mahendran If unsure, say N. 7310f050d99SGanesh Mahendran 7329e5c33d7SMark Salterconfig GENERIC_EARLY_IOREMAP 7339e5c33d7SMark Salter bool 734042d27acSHelge Deller 735042d27acSHelge Dellerconfig MAX_STACK_SIZE_MB 736042d27acSHelge Deller int "Maximum user stack size for 32-bit processes (MB)" 737042d27acSHelge Deller default 80 738042d27acSHelge Deller range 8 2048 739042d27acSHelge Deller depends on STACK_GROWSUP && (!64BIT || COMPAT) 740042d27acSHelge Deller help 741042d27acSHelge Deller This is the maximum stack size in Megabytes in the VM layout of 32-bit 742042d27acSHelge Deller user processes when the stack grows upwards (currently only on parisc 7435f171577SJames Hogan arch). The stack will be located at the highest memory address minus 7445f171577SJames Hogan the given value, unless the RLIMIT_STACK hard limit is changed to a 7455f171577SJames Hogan smaller value in which case that is used. 746042d27acSHelge Deller 747042d27acSHelge Deller A sane initial value is 80 MB. 7483a80a7faSMel Gorman 7493a80a7faSMel Gormanconfig DEFERRED_STRUCT_PAGE_INIT 7501ce22103SVlastimil Babka bool "Defer initialisation of struct pages to kthreads" 751d39f8fb4SMike Rapoport depends on SPARSEMEM 752ab1e8d89SPavel Tatashin depends on !NEED_PER_CPU_KM 753889c695dSPasha Tatashin depends on 64BIT 754e4443149SDaniel Jordan select PADATA 7553a80a7faSMel Gorman help 7563a80a7faSMel Gorman Ordinarily all struct pages are initialised during early boot in a 7573a80a7faSMel Gorman single thread. On very large machines this can take a considerable 7583a80a7faSMel Gorman amount of time. If this option is set, large machines will bring up 759e4443149SDaniel Jordan a subset of memmap at boot and then initialise the rest in parallel. 760e4443149SDaniel Jordan This has a potential performance impact on tasks running early in the 7611ce22103SVlastimil Babka lifetime of the system until these kthreads finish the 7621ce22103SVlastimil Babka initialisation. 763033fbae9SDan Williams 76433c3fc71SVladimir Davydovconfig IDLE_PAGE_TRACKING 76533c3fc71SVladimir Davydov bool "Enable idle page tracking" 76633c3fc71SVladimir Davydov depends on SYSFS && MMU 76733c3fc71SVladimir Davydov select PAGE_EXTENSION if !64BIT 76833c3fc71SVladimir Davydov help 76933c3fc71SVladimir Davydov This feature allows to estimate the amount of user pages that have 77033c3fc71SVladimir Davydov not been touched during a given period of time. This information can 77133c3fc71SVladimir Davydov be useful to tune memory cgroup limits and/or for job placement 77233c3fc71SVladimir Davydov within a compute cluster. 77333c3fc71SVladimir Davydov 7741ad1335dSMike Rapoport See Documentation/admin-guide/mm/idle_page_tracking.rst for 7751ad1335dSMike Rapoport more details. 77633c3fc71SVladimir Davydov 77717596731SRobin Murphyconfig ARCH_HAS_PTE_DEVMAP 77865f7d049SOliver O'Halloran bool 77965f7d049SOliver O'Halloran 780033fbae9SDan Williamsconfig ZONE_DEVICE 7815042db43SJérôme Glisse bool "Device memory (pmem, HMM, etc...) hotplug support" 782033fbae9SDan Williams depends on MEMORY_HOTPLUG 783033fbae9SDan Williams depends on MEMORY_HOTREMOVE 78499490f16SDan Williams depends on SPARSEMEM_VMEMMAP 78517596731SRobin Murphy depends on ARCH_HAS_PTE_DEVMAP 7863a08cd52SMatthew Wilcox select XARRAY_MULTI 787033fbae9SDan Williams 788033fbae9SDan Williams help 789033fbae9SDan Williams Device memory hotplug support allows for establishing pmem, 790033fbae9SDan Williams or other device driver discovered memory regions, in the 791033fbae9SDan Williams memmap. This allows pfn_to_page() lookups of otherwise 792033fbae9SDan Williams "device-physical" addresses which is needed for using a DAX 793033fbae9SDan Williams mapping in an O_DIRECT operation, among other things. 794033fbae9SDan Williams 795033fbae9SDan Williams If FS_DAX is enabled, then say Y. 79606a660adSLinus Torvalds 797e7638488SDan Williamsconfig DEV_PAGEMAP_OPS 798e7638488SDan Williams bool 799e7638488SDan Williams 8009c240a7bSChristoph Hellwig# 8019c240a7bSChristoph Hellwig# Helpers to mirror range of the CPU page tables of a process into device page 8029c240a7bSChristoph Hellwig# tables. 8039c240a7bSChristoph Hellwig# 804c0b12405SJérôme Glisseconfig HMM_MIRROR 8059c240a7bSChristoph Hellwig bool 806f442c283SChristoph Hellwig depends on MMU 807c0b12405SJérôme Glisse 8085042db43SJérôme Glisseconfig DEVICE_PRIVATE 8095042db43SJérôme Glisse bool "Unaddressable device memory (GPU memory, ...)" 8107328d9ccSChristoph Hellwig depends on ZONE_DEVICE 811e7638488SDan Williams select DEV_PAGEMAP_OPS 8125042db43SJérôme Glisse 8135042db43SJérôme Glisse help 8145042db43SJérôme Glisse Allows creation of struct pages to represent unaddressable device 8155042db43SJérôme Glisse memory; i.e., memory that is only accessible from the device (or 8165042db43SJérôme Glisse group of devices). You likely also want to select HMM_MIRROR. 8175042db43SJérôme Glisse 8188025e5ddSJan Karaconfig FRAME_VECTOR 8198025e5ddSJan Kara bool 82063c17fb8SDave Hansen 82163c17fb8SDave Hansenconfig ARCH_USES_HIGH_VMA_FLAGS 82263c17fb8SDave Hansen bool 82366d37570SDave Hansenconfig ARCH_HAS_PKEYS 82466d37570SDave Hansen bool 82530a5b536SDennis Zhou 82630a5b536SDennis Zhouconfig PERCPU_STATS 82730a5b536SDennis Zhou bool "Collect percpu memory statistics" 82830a5b536SDennis Zhou help 82930a5b536SDennis Zhou This feature collects and exposes statistics via debugfs. The 83030a5b536SDennis Zhou information includes global and per chunk statistics, which can 83130a5b536SDennis Zhou be used to help understand percpu memory usage. 83264c349f4SKirill A. Shutemov 83364c349f4SKirill A. Shutemovconfig GUP_BENCHMARK 83464c349f4SKirill A. Shutemov bool "Enable infrastructure for get_user_pages_fast() benchmarking" 83564c349f4SKirill A. Shutemov help 83664c349f4SKirill A. Shutemov Provides /sys/kernel/debug/gup_benchmark that helps with testing 83764c349f4SKirill A. Shutemov performance of get_user_pages_fast(). 83864c349f4SKirill A. Shutemov 83964c349f4SKirill A. Shutemov See tools/testing/selftests/vm/gup_benchmark.c 8403010a5eaSLaurent Dufour 84139656e83SChristoph Hellwigconfig GUP_GET_PTE_LOW_HIGH 84239656e83SChristoph Hellwig bool 84339656e83SChristoph Hellwig 84499cb0dbdSSong Liuconfig READ_ONLY_THP_FOR_FS 84599cb0dbdSSong Liu bool "Read-only THP for filesystems (EXPERIMENTAL)" 846396bcc52SMatthew Wilcox (Oracle) depends on TRANSPARENT_HUGEPAGE && SHMEM 84799cb0dbdSSong Liu 84899cb0dbdSSong Liu help 84999cb0dbdSSong Liu Allow khugepaged to put read-only file-backed pages in THP. 85099cb0dbdSSong Liu 85199cb0dbdSSong Liu This is marked experimental because it is a new feature. Write 85299cb0dbdSSong Liu support of file THPs will be developed in the next few release 85399cb0dbdSSong Liu cycles. 85499cb0dbdSSong Liu 8553010a5eaSLaurent Dufourconfig ARCH_HAS_PTE_SPECIAL 8563010a5eaSLaurent Dufour bool 85759e0b520SChristoph Hellwig 858cbd34da7SChristoph Hellwig# 859cbd34da7SChristoph Hellwig# Some architectures require a special hugepage directory format that is 860cbd34da7SChristoph Hellwig# required to support multiple hugepage sizes. For example a4fe3ce76 861cbd34da7SChristoph Hellwig# "powerpc/mm: Allow more flexible layouts for hugepage pagetables" 862cbd34da7SChristoph Hellwig# introduced it on powerpc. This allows for a more flexible hugepage 863cbd34da7SChristoph Hellwig# pagetable layouts. 864cbd34da7SChristoph Hellwig# 865cbd34da7SChristoph Hellwigconfig ARCH_HAS_HUGEPD 866cbd34da7SChristoph Hellwig bool 867cbd34da7SChristoph Hellwig 868c5acad84SThomas Hellstromconfig MAPPING_DIRTY_HELPERS 869c5acad84SThomas Hellstrom bool 870c5acad84SThomas Hellstrom 87159e0b520SChristoph Hellwigendmenu 872