1e1785e85SDave Hansenconfig SELECT_MEMORY_MODEL 2e1785e85SDave Hansen def_bool y 3a8826eebSKees Cook depends on ARCH_SELECT_MEMORY_MODEL 4e1785e85SDave Hansen 53a9da765SDave Hansenchoice 63a9da765SDave Hansen prompt "Memory model" 7e1785e85SDave Hansen depends on SELECT_MEMORY_MODEL 8e1785e85SDave Hansen default DISCONTIGMEM_MANUAL if ARCH_DISCONTIGMEM_DEFAULT 9d41dee36SAndy Whitcroft default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT 10e1785e85SDave Hansen default FLATMEM_MANUAL 113a9da765SDave Hansen 12e1785e85SDave Hansenconfig FLATMEM_MANUAL 133a9da765SDave Hansen bool "Flat Memory" 14c898ec16SAnton Blanchard depends on !(ARCH_DISCONTIGMEM_ENABLE || ARCH_SPARSEMEM_ENABLE) || ARCH_FLATMEM_ENABLE 153a9da765SDave Hansen help 163a9da765SDave Hansen This option allows you to change some of the ways that 173a9da765SDave Hansen Linux manages its memory internally. Most users will 183a9da765SDave Hansen only have one option here: FLATMEM. This is normal 193a9da765SDave Hansen and a correct option. 203a9da765SDave Hansen 21d41dee36SAndy Whitcroft Some users of more advanced features like NUMA and 22d41dee36SAndy Whitcroft memory hotplug may have different options here. 2318f65332SGeert Uytterhoeven DISCONTIGMEM is a more mature, better tested system, 24d41dee36SAndy Whitcroft but is incompatible with memory hotplug and may suffer 25d41dee36SAndy Whitcroft decreased performance over SPARSEMEM. If unsure between 26d41dee36SAndy Whitcroft "Sparse Memory" and "Discontiguous Memory", choose 27d41dee36SAndy Whitcroft "Discontiguous Memory". 28d41dee36SAndy Whitcroft 29d41dee36SAndy Whitcroft If unsure, choose this option (Flat Memory) over any other. 303a9da765SDave Hansen 31e1785e85SDave Hansenconfig DISCONTIGMEM_MANUAL 32f3519f91SDave Hansen bool "Discontiguous Memory" 333a9da765SDave Hansen depends on ARCH_DISCONTIGMEM_ENABLE 343a9da765SDave Hansen help 35785dcd44SDave Hansen This option provides enhanced support for discontiguous 36785dcd44SDave Hansen memory systems, over FLATMEM. These systems have holes 37785dcd44SDave Hansen in their physical address spaces, and this option provides 38785dcd44SDave Hansen more efficient handling of these holes. However, the vast 39785dcd44SDave Hansen majority of hardware has quite flat address spaces, and 40ad3d0a38SPhilipp Marek can have degraded performance from the extra overhead that 41785dcd44SDave Hansen this option imposes. 42785dcd44SDave Hansen 43785dcd44SDave Hansen Many NUMA configurations will have this as the only option. 44785dcd44SDave Hansen 453a9da765SDave Hansen If unsure, choose "Flat Memory" over this option. 463a9da765SDave Hansen 47d41dee36SAndy Whitcroftconfig SPARSEMEM_MANUAL 48d41dee36SAndy Whitcroft bool "Sparse Memory" 49d41dee36SAndy Whitcroft depends on ARCH_SPARSEMEM_ENABLE 50d41dee36SAndy Whitcroft help 51d41dee36SAndy Whitcroft This will be the only option for some systems, including 52d41dee36SAndy Whitcroft memory hotplug systems. This is normal. 53d41dee36SAndy Whitcroft 54d41dee36SAndy Whitcroft For many other systems, this will be an alternative to 55f3519f91SDave Hansen "Discontiguous Memory". This option provides some potential 56d41dee36SAndy Whitcroft performance benefits, along with decreased code complexity, 57d41dee36SAndy Whitcroft but it is newer, and more experimental. 58d41dee36SAndy Whitcroft 59d41dee36SAndy Whitcroft If unsure, choose "Discontiguous Memory" or "Flat Memory" 60d41dee36SAndy Whitcroft over this option. 61d41dee36SAndy Whitcroft 623a9da765SDave Hansenendchoice 633a9da765SDave Hansen 64e1785e85SDave Hansenconfig DISCONTIGMEM 65e1785e85SDave Hansen def_bool y 66e1785e85SDave Hansen depends on (!SELECT_MEMORY_MODEL && ARCH_DISCONTIGMEM_ENABLE) || DISCONTIGMEM_MANUAL 67e1785e85SDave Hansen 68d41dee36SAndy Whitcroftconfig SPARSEMEM 69d41dee36SAndy Whitcroft def_bool y 701a83e175SRussell King depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL 71d41dee36SAndy Whitcroft 72e1785e85SDave Hansenconfig FLATMEM 73e1785e85SDave Hansen def_bool y 74d41dee36SAndy Whitcroft depends on (!DISCONTIGMEM && !SPARSEMEM) || FLATMEM_MANUAL 75d41dee36SAndy Whitcroft 76d41dee36SAndy Whitcroftconfig FLAT_NODE_MEM_MAP 77d41dee36SAndy Whitcroft def_bool y 78d41dee36SAndy Whitcroft depends on !SPARSEMEM 79e1785e85SDave Hansen 8093b7504eSDave Hansen# 8193b7504eSDave Hansen# Both the NUMA code and DISCONTIGMEM use arrays of pg_data_t's 8293b7504eSDave Hansen# to represent different areas of memory. This variable allows 8393b7504eSDave Hansen# those dependencies to exist individually. 8493b7504eSDave Hansen# 8593b7504eSDave Hansenconfig NEED_MULTIPLE_NODES 8693b7504eSDave Hansen def_bool y 8793b7504eSDave Hansen depends on DISCONTIGMEM || NUMA 88af705362SAndy Whitcroft 89af705362SAndy Whitcroftconfig HAVE_MEMORY_PRESENT 90af705362SAndy Whitcroft def_bool y 91d41dee36SAndy Whitcroft depends on ARCH_HAVE_MEMORY_PRESENT || SPARSEMEM 92802f192eSBob Picco 93802f192eSBob Picco# 943e347261SBob Picco# SPARSEMEM_EXTREME (which is the default) does some bootmem 953e347261SBob Picco# allocations when memory_present() is called. If this cannot 963e347261SBob Picco# be done on your architecture, select this option. However, 973e347261SBob Picco# statically allocating the mem_section[] array can potentially 983e347261SBob Picco# consume vast quantities of .bss, so be careful. 993e347261SBob Picco# 1003e347261SBob Picco# This option will also potentially produce smaller runtime code 1013e347261SBob Picco# with gcc 3.4 and later. 1023e347261SBob Picco# 1033e347261SBob Piccoconfig SPARSEMEM_STATIC 1049ba16087SJan Beulich bool 1053e347261SBob Picco 1063e347261SBob Picco# 10744c09201SMatt LaPlante# Architecture platforms which require a two level mem_section in SPARSEMEM 108802f192eSBob Picco# must select this option. This is usually for architecture platforms with 109802f192eSBob Picco# an extremely sparse physical address space. 110802f192eSBob Picco# 1113e347261SBob Piccoconfig SPARSEMEM_EXTREME 1123e347261SBob Picco def_bool y 1133e347261SBob Picco depends on SPARSEMEM && !SPARSEMEM_STATIC 1144c21e2f2SHugh Dickins 11529c71111SAndy Whitcroftconfig SPARSEMEM_VMEMMAP_ENABLE 1169ba16087SJan Beulich bool 11729c71111SAndy Whitcroft 1189bdac914SYinghai Luconfig SPARSEMEM_ALLOC_MEM_MAP_TOGETHER 1199bdac914SYinghai Lu def_bool y 1209bdac914SYinghai Lu depends on SPARSEMEM && X86_64 1219bdac914SYinghai Lu 12229c71111SAndy Whitcroftconfig SPARSEMEM_VMEMMAP 123a5ee6daaSGeoff Levand bool "Sparse Memory virtual memmap" 124a5ee6daaSGeoff Levand depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE 125a5ee6daaSGeoff Levand default y 126a5ee6daaSGeoff Levand help 127a5ee6daaSGeoff Levand SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise 128a5ee6daaSGeoff Levand pfn_to_page and page_to_pfn operations. This is the most 129a5ee6daaSGeoff Levand efficient option when sufficient kernel resources are available. 13029c71111SAndy Whitcroft 13195f72d1eSYinghai Luconfig HAVE_MEMBLOCK 13295f72d1eSYinghai Lu boolean 13395f72d1eSYinghai Lu 1347c0caeb8STejun Heoconfig HAVE_MEMBLOCK_NODE_MAP 1357c0caeb8STejun Heo boolean 1367c0caeb8STejun Heo 137c378ddd5STejun Heoconfig ARCH_DISCARD_MEMBLOCK 138c378ddd5STejun Heo boolean 139c378ddd5STejun Heo 14066616720SSam Ravnborgconfig NO_BOOTMEM 14166616720SSam Ravnborg boolean 14266616720SSam Ravnborg 143ee6f509cSMinchan Kimconfig MEMORY_ISOLATION 144ee6f509cSMinchan Kim boolean 145ee6f509cSMinchan Kim 14620b2f52bSLai Jiangshanconfig MOVABLE_NODE 14720b2f52bSLai Jiangshan boolean "Enable to assign a node which has only movable memory" 14820b2f52bSLai Jiangshan depends on HAVE_MEMBLOCK 14920b2f52bSLai Jiangshan depends on NO_BOOTMEM 15020b2f52bSLai Jiangshan depends on X86_64 15120b2f52bSLai Jiangshan depends on NUMA 152c2974058STang Chen default n 153c2974058STang Chen help 154c2974058STang Chen Allow a node to have only movable memory. Pages used by the kernel, 155c2974058STang Chen such as direct mapping pages cannot be migrated. So the corresponding 156c5320926STang Chen memory device cannot be hotplugged. This option allows the following 157c5320926STang Chen two things: 158c5320926STang Chen - When the system is booting, node full of hotpluggable memory can 159c5320926STang Chen be arranged to have only movable memory so that the whole node can 160c5320926STang Chen be hot-removed. (need movable_node boot option specified). 161c5320926STang Chen - After the system is up, the option allows users to online all the 162c5320926STang Chen memory of a node as movable memory so that the whole node can be 163c5320926STang Chen hot-removed. 164c5320926STang Chen 165c5320926STang Chen Users who don't use the memory hotplug feature are fine with this 166c5320926STang Chen option on since they don't specify movable_node boot option or they 167c5320926STang Chen don't online memory as movable. 168c2974058STang Chen 169c2974058STang Chen Say Y here if you want to hotplug a whole node. 170c2974058STang Chen Say N here if you want kernel to use memory on all nodes evenly. 17120b2f52bSLai Jiangshan 17246723bfaSYasuaki Ishimatsu# 17346723bfaSYasuaki Ishimatsu# Only be set on architectures that have completely implemented memory hotplug 17446723bfaSYasuaki Ishimatsu# feature. If you are not sure, don't touch it. 17546723bfaSYasuaki Ishimatsu# 17646723bfaSYasuaki Ishimatsuconfig HAVE_BOOTMEM_INFO_NODE 17746723bfaSYasuaki Ishimatsu def_bool n 17846723bfaSYasuaki Ishimatsu 1793947be19SDave Hansen# eventually, we can have this option just 'select SPARSEMEM' 1803947be19SDave Hansenconfig MEMORY_HOTPLUG 1813947be19SDave Hansen bool "Allow for memory hot-add" 182ec69acbbSKeith Mannthey depends on SPARSEMEM || X86_64_ACPI_NUMA 18340b31360SStephen Rothwell depends on ARCH_ENABLE_MEMORY_HOTPLUG 184ed84a07aSKumar Gala depends on (IA64 || X86 || PPC_BOOK3S_64 || SUPERH || S390) 1853947be19SDave Hansen 186ec69acbbSKeith Manntheyconfig MEMORY_HOTPLUG_SPARSE 187ec69acbbSKeith Mannthey def_bool y 188ec69acbbSKeith Mannthey depends on SPARSEMEM && MEMORY_HOTPLUG 189ec69acbbSKeith Mannthey 1900c0e6195SKAMEZAWA Hiroyukiconfig MEMORY_HOTREMOVE 1910c0e6195SKAMEZAWA Hiroyuki bool "Allow for memory hot remove" 19246723bfaSYasuaki Ishimatsu select MEMORY_ISOLATION 193f7e3334aSNathan Fontenot select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64) 1940c0e6195SKAMEZAWA Hiroyuki depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE 1950c0e6195SKAMEZAWA Hiroyuki depends on MIGRATION 1960c0e6195SKAMEZAWA Hiroyuki 197e20b8ccaSChristoph Lameter# 198e20b8ccaSChristoph Lameter# If we have space for more page flags then we can enable additional 199e20b8ccaSChristoph Lameter# optimizations and functionality. 200e20b8ccaSChristoph Lameter# 201e20b8ccaSChristoph Lameter# Regular Sparsemem takes page flag bits for the sectionid if it does not 202e20b8ccaSChristoph Lameter# use a virtual memmap. Disable extended page flags for 32 bit platforms 203e20b8ccaSChristoph Lameter# that require the use of a sectionid in the page flags. 204e20b8ccaSChristoph Lameter# 205e20b8ccaSChristoph Lameterconfig PAGEFLAGS_EXTENDED 206e20b8ccaSChristoph Lameter def_bool y 207a269cca9SH. Peter Anvin depends on 64BIT || SPARSEMEM_VMEMMAP || !SPARSEMEM 208e20b8ccaSChristoph Lameter 2094c21e2f2SHugh Dickins# Heavily threaded applications may benefit from splitting the mm-wide 2104c21e2f2SHugh Dickins# page_table_lock, so that faults on different parts of the user address 2114c21e2f2SHugh Dickins# space can be handled with less contention: split it at this NR_CPUS. 2124c21e2f2SHugh Dickins# Default to 4 for wider testing, though 8 might be more appropriate. 2134c21e2f2SHugh Dickins# ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock. 2147b6ac9dfSHugh Dickins# PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes. 215a70caa8bSHugh Dickins# DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page. 2164c21e2f2SHugh Dickins# 2174c21e2f2SHugh Dickinsconfig SPLIT_PTLOCK_CPUS 2184c21e2f2SHugh Dickins int 219a70caa8bSHugh Dickins default "999999" if ARM && !CPU_CACHE_VIPT 220a70caa8bSHugh Dickins default "999999" if PARISC && !PA20 2214c21e2f2SHugh Dickins default "4" 2227cbe34cfSChristoph Lameter 223e009bb30SKirill A. Shutemovconfig ARCH_ENABLE_SPLIT_PMD_PTLOCK 224e009bb30SKirill A. Shutemov boolean 225e009bb30SKirill A. Shutemov 2267cbe34cfSChristoph Lameter# 22718468d93SRafael Aquini# support for memory balloon compaction 22818468d93SRafael Aquiniconfig BALLOON_COMPACTION 22918468d93SRafael Aquini bool "Allow for balloon memory compaction/migration" 23018468d93SRafael Aquini def_bool y 23118468d93SRafael Aquini depends on COMPACTION && VIRTIO_BALLOON 23218468d93SRafael Aquini help 23318468d93SRafael Aquini Memory fragmentation introduced by ballooning might reduce 23418468d93SRafael Aquini significantly the number of 2MB contiguous memory blocks that can be 23518468d93SRafael Aquini used within a guest, thus imposing performance penalties associated 23618468d93SRafael Aquini with the reduced number of transparent huge pages that could be used 23718468d93SRafael Aquini by the guest workload. Allowing the compaction & migration for memory 23818468d93SRafael Aquini pages enlisted as being part of memory balloon devices avoids the 23918468d93SRafael Aquini scenario aforementioned and helps improving memory defragmentation. 24018468d93SRafael Aquini 24118468d93SRafael Aquini# 242e9e96b39SMel Gorman# support for memory compaction 243e9e96b39SMel Gormanconfig COMPACTION 244e9e96b39SMel Gorman bool "Allow for memory compaction" 24505106e6aSRik van Riel def_bool y 246e9e96b39SMel Gorman select MIGRATION 24733a93877SAndrea Arcangeli depends on MMU 248e9e96b39SMel Gorman help 249e9e96b39SMel Gorman Allows the compaction of memory for the allocation of huge pages. 250e9e96b39SMel Gorman 251e9e96b39SMel Gorman# 2527cbe34cfSChristoph Lameter# support for page migration 2537cbe34cfSChristoph Lameter# 2547cbe34cfSChristoph Lameterconfig MIGRATION 255b20a3503SChristoph Lameter bool "Page migration" 2566c5240aeSChristoph Lameter def_bool y 257de32a817SChen Gang depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU 258b20a3503SChristoph Lameter help 259b20a3503SChristoph Lameter Allows the migration of the physical location of pages of processes 260e9e96b39SMel Gorman while the virtual addresses are not changed. This is useful in 261e9e96b39SMel Gorman two situations. The first is on NUMA systems to put pages nearer 262e9e96b39SMel Gorman to the processors accessing. The second is when allocating huge 263e9e96b39SMel Gorman pages as migration can relocate pages to satisfy a huge page 264e9e96b39SMel Gorman allocation instead of reclaiming. 2656550e07fSGreg Kroah-Hartman 266600715dcSJeremy Fitzhardingeconfig PHYS_ADDR_T_64BIT 267600715dcSJeremy Fitzhardinge def_bool 64BIT || ARCH_PHYS_ADDR_T_64BIT 268600715dcSJeremy Fitzhardinge 2694b51d669SChristoph Lameterconfig ZONE_DMA_FLAG 2704b51d669SChristoph Lameter int 2714b51d669SChristoph Lameter default "0" if !ZONE_DMA 2724b51d669SChristoph Lameter default "1" 2734b51d669SChristoph Lameter 2742a7326b5SChristoph Lameterconfig BOUNCE 2759ca24e2eSVinayak Menon bool "Enable bounce buffers" 2769ca24e2eSVinayak Menon default y 2772a7326b5SChristoph Lameter depends on BLOCK && MMU && (ZONE_DMA || HIGHMEM) 2789ca24e2eSVinayak Menon help 2799ca24e2eSVinayak Menon Enable bounce buffers for devices that cannot access 2809ca24e2eSVinayak Menon the full range of memory available to the CPU. Enabled 2819ca24e2eSVinayak Menon by default when ZONE_DMA or HIGHMEM is selected, but you 2829ca24e2eSVinayak Menon may say n to override this. 2832a7326b5SChristoph Lameter 284ffecfd1aSDarrick J. Wong# On the 'tile' arch, USB OHCI needs the bounce pool since tilegx will often 285ffecfd1aSDarrick J. Wong# have more than 4GB of memory, but we don't currently use the IOTLB to present 286ffecfd1aSDarrick J. Wong# a 32-bit address to OHCI. So we need to use a bounce pool instead. 287ffecfd1aSDarrick J. Wong# 288ffecfd1aSDarrick J. Wong# We also use the bounce pool to provide stable page writes for jbd. jbd 289ffecfd1aSDarrick J. Wong# initiates buffer writeback without locking the page or setting PG_writeback, 290ffecfd1aSDarrick J. Wong# and fixing that behavior (a second time; jbd2 doesn't have this problem) is 291ffecfd1aSDarrick J. Wong# a major rework effort. Instead, use the bounce buffer to snapshot pages 292ffecfd1aSDarrick J. Wong# (until jbd goes away). The only jbd user is ext3. 293ffecfd1aSDarrick J. Wongconfig NEED_BOUNCE_POOL 294ffecfd1aSDarrick J. Wong bool 295ffecfd1aSDarrick J. Wong default y if (TILE && USB_OHCI_HCD) || (BLK_DEV_INTEGRITY && JBD) 296ffecfd1aSDarrick J. Wong 2976225e937SChristoph Lameterconfig NR_QUICK 2986225e937SChristoph Lameter int 2996225e937SChristoph Lameter depends on QUICKLIST 3000176bd3dSPaul Mundt default "2" if AVR32 3016225e937SChristoph Lameter default "1" 302f057eac0SStephen Rothwell 303f057eac0SStephen Rothwellconfig VIRT_TO_BUS 3044febd95aSStephen Rothwell bool 3054febd95aSStephen Rothwell help 3064febd95aSStephen Rothwell An architecture should select this if it implements the 3074febd95aSStephen Rothwell deprecated interface virt_to_bus(). All new architectures 3084febd95aSStephen Rothwell should probably not select this. 3094febd95aSStephen Rothwell 310cddb8a5cSAndrea Arcangeli 311cddb8a5cSAndrea Arcangeliconfig MMU_NOTIFIER 312cddb8a5cSAndrea Arcangeli bool 313fc4d5c29SDavid Howells 314f8af4da3SHugh Dickinsconfig KSM 315f8af4da3SHugh Dickins bool "Enable KSM for page merging" 316f8af4da3SHugh Dickins depends on MMU 317f8af4da3SHugh Dickins help 318f8af4da3SHugh Dickins Enable Kernel Samepage Merging: KSM periodically scans those areas 319f8af4da3SHugh Dickins of an application's address space that an app has advised may be 320f8af4da3SHugh Dickins mergeable. When it finds pages of identical content, it replaces 321d0f209f6SHugh Dickins the many instances by a single page with that content, so 322f8af4da3SHugh Dickins saving memory until one or another app needs to modify the content. 323f8af4da3SHugh Dickins Recommended for use with KVM, or with other duplicative applications. 324c73602adSHugh Dickins See Documentation/vm/ksm.txt for more information: KSM is inactive 325c73602adSHugh Dickins until a program has madvised that an area is MADV_MERGEABLE, and 326c73602adSHugh Dickins root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). 327f8af4da3SHugh Dickins 328e0a94c2aSChristoph Lameterconfig DEFAULT_MMAP_MIN_ADDR 329e0a94c2aSChristoph Lameter int "Low address space to protect from user allocation" 3306e141546SDavid Howells depends on MMU 331e0a94c2aSChristoph Lameter default 4096 332e0a94c2aSChristoph Lameter help 333e0a94c2aSChristoph Lameter This is the portion of low virtual memory which should be protected 334e0a94c2aSChristoph Lameter from userspace allocation. Keeping a user from writing to low pages 335e0a94c2aSChristoph Lameter can help reduce the impact of kernel NULL pointer bugs. 336e0a94c2aSChristoph Lameter 337e0a94c2aSChristoph Lameter For most ia64, ppc64 and x86 users with lots of address space 338e0a94c2aSChristoph Lameter a value of 65536 is reasonable and should cause no problems. 339e0a94c2aSChristoph Lameter On arm and other archs it should not be higher than 32768. 340788084abSEric Paris Programs which use vm86 functionality or have some need to map 341788084abSEric Paris this low address space will need CAP_SYS_RAWIO or disable this 342788084abSEric Paris protection by setting the value to 0. 343e0a94c2aSChristoph Lameter 344e0a94c2aSChristoph Lameter This value can be changed after boot using the 345e0a94c2aSChristoph Lameter /proc/sys/vm/mmap_min_addr tunable. 346e0a94c2aSChristoph Lameter 347d949f36fSLinus Torvaldsconfig ARCH_SUPPORTS_MEMORY_FAILURE 348d949f36fSLinus Torvalds bool 349e0a94c2aSChristoph Lameter 3506a46079cSAndi Kleenconfig MEMORY_FAILURE 3516a46079cSAndi Kleen depends on MMU 352d949f36fSLinus Torvalds depends on ARCH_SUPPORTS_MEMORY_FAILURE 3536a46079cSAndi Kleen bool "Enable recovery from hardware memory errors" 354ee6f509cSMinchan Kim select MEMORY_ISOLATION 3556a46079cSAndi Kleen help 3566a46079cSAndi Kleen Enables code to recover from some memory failures on systems 3576a46079cSAndi Kleen with MCA recovery. This allows a system to continue running 3586a46079cSAndi Kleen even when some of its memory has uncorrected errors. This requires 3596a46079cSAndi Kleen special hardware support and typically ECC memory. 3606a46079cSAndi Kleen 361cae681fcSAndi Kleenconfig HWPOISON_INJECT 362413f9efbSAndi Kleen tristate "HWPoison pages injector" 36327df5068SAndi Kleen depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS 364478c5ffcSWu Fengguang select PROC_PAGE_MONITOR 365cae681fcSAndi Kleen 366fc4d5c29SDavid Howellsconfig NOMMU_INITIAL_TRIM_EXCESS 367fc4d5c29SDavid Howells int "Turn on mmap() excess space trimming before booting" 368fc4d5c29SDavid Howells depends on !MMU 369fc4d5c29SDavid Howells default 1 370fc4d5c29SDavid Howells help 371fc4d5c29SDavid Howells The NOMMU mmap() frequently needs to allocate large contiguous chunks 372fc4d5c29SDavid Howells of memory on which to store mappings, but it can only ask the system 373fc4d5c29SDavid Howells allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently 374fc4d5c29SDavid Howells more than it requires. To deal with this, mmap() is able to trim off 375fc4d5c29SDavid Howells the excess and return it to the allocator. 376fc4d5c29SDavid Howells 377fc4d5c29SDavid Howells If trimming is enabled, the excess is trimmed off and returned to the 378fc4d5c29SDavid Howells system allocator, which can cause extra fragmentation, particularly 379fc4d5c29SDavid Howells if there are a lot of transient processes. 380fc4d5c29SDavid Howells 381fc4d5c29SDavid Howells If trimming is disabled, the excess is kept, but not used, which for 382fc4d5c29SDavid Howells long-term mappings means that the space is wasted. 383fc4d5c29SDavid Howells 384fc4d5c29SDavid Howells Trimming can be dynamically controlled through a sysctl option 385fc4d5c29SDavid Howells (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of 386fc4d5c29SDavid Howells excess pages there must be before trimming should occur, or zero if 387fc4d5c29SDavid Howells no trimming is to occur. 388fc4d5c29SDavid Howells 389fc4d5c29SDavid Howells This option specifies the initial value of this option. The default 390fc4d5c29SDavid Howells of 1 says that all excess pages should be trimmed. 391fc4d5c29SDavid Howells 392fc4d5c29SDavid Howells See Documentation/nommu-mmap.txt for more information. 393bbddff05STejun Heo 3944c76d9d1SAndrea Arcangeliconfig TRANSPARENT_HUGEPAGE 39513ece886SAndrea Arcangeli bool "Transparent Hugepage Support" 39615626062SGerald Schaefer depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE 3975d689240SAndrea Arcangeli select COMPACTION 3984c76d9d1SAndrea Arcangeli help 3994c76d9d1SAndrea Arcangeli Transparent Hugepages allows the kernel to use huge pages and 4004c76d9d1SAndrea Arcangeli huge tlb transparently to the applications whenever possible. 4014c76d9d1SAndrea Arcangeli This feature can improve computing performance to certain 4024c76d9d1SAndrea Arcangeli applications by speeding up page faults during memory 4034c76d9d1SAndrea Arcangeli allocation, by reducing the number of tlb misses and by speeding 4044c76d9d1SAndrea Arcangeli up the pagetable walking. 4054c76d9d1SAndrea Arcangeli 4064c76d9d1SAndrea Arcangeli If memory constrained on embedded, you may want to say N. 4074c76d9d1SAndrea Arcangeli 40813ece886SAndrea Arcangelichoice 40913ece886SAndrea Arcangeli prompt "Transparent Hugepage Support sysfs defaults" 41013ece886SAndrea Arcangeli depends on TRANSPARENT_HUGEPAGE 41113ece886SAndrea Arcangeli default TRANSPARENT_HUGEPAGE_ALWAYS 41213ece886SAndrea Arcangeli help 41313ece886SAndrea Arcangeli Selects the sysfs defaults for Transparent Hugepage Support. 41413ece886SAndrea Arcangeli 41513ece886SAndrea Arcangeli config TRANSPARENT_HUGEPAGE_ALWAYS 41613ece886SAndrea Arcangeli bool "always" 41713ece886SAndrea Arcangeli help 41813ece886SAndrea Arcangeli Enabling Transparent Hugepage always, can increase the 41913ece886SAndrea Arcangeli memory footprint of applications without a guaranteed 42013ece886SAndrea Arcangeli benefit but it will work automatically for all applications. 42113ece886SAndrea Arcangeli 42213ece886SAndrea Arcangeli config TRANSPARENT_HUGEPAGE_MADVISE 42313ece886SAndrea Arcangeli bool "madvise" 42413ece886SAndrea Arcangeli help 42513ece886SAndrea Arcangeli Enabling Transparent Hugepage madvise, will only provide a 42613ece886SAndrea Arcangeli performance improvement benefit to the applications using 42713ece886SAndrea Arcangeli madvise(MADV_HUGEPAGE) but it won't risk to increase the 42813ece886SAndrea Arcangeli memory footprint of applications without a guaranteed 42913ece886SAndrea Arcangeli benefit. 43013ece886SAndrea Arcangeliendchoice 43113ece886SAndrea Arcangeli 4325febcbe9SChristopher Yeohconfig CROSS_MEMORY_ATTACH 4335febcbe9SChristopher Yeoh bool "Cross Memory Support" 4345febcbe9SChristopher Yeoh depends on MMU 4355febcbe9SChristopher Yeoh default y 4365febcbe9SChristopher Yeoh help 4375febcbe9SChristopher Yeoh Enabling this option adds the system calls process_vm_readv and 4385febcbe9SChristopher Yeoh process_vm_writev which allow a process with the correct privileges 4395febcbe9SChristopher Yeoh to directly read from or write to to another process's address space. 4405febcbe9SChristopher Yeoh See the man page for more details. 4415febcbe9SChristopher Yeoh 442bbddff05STejun Heo# 443bbddff05STejun Heo# UP and nommu archs use km based percpu allocator 444bbddff05STejun Heo# 445bbddff05STejun Heoconfig NEED_PER_CPU_KM 446bbddff05STejun Heo depends on !SMP 447bbddff05STejun Heo bool 448bbddff05STejun Heo default y 449077b1f83SDan Magenheimer 450077b1f83SDan Magenheimerconfig CLEANCACHE 451077b1f83SDan Magenheimer bool "Enable cleancache driver to cache clean pages if tmem is present" 452077b1f83SDan Magenheimer default n 453077b1f83SDan Magenheimer help 454077b1f83SDan Magenheimer Cleancache can be thought of as a page-granularity victim cache 455077b1f83SDan Magenheimer for clean pages that the kernel's pageframe replacement algorithm 456077b1f83SDan Magenheimer (PFRA) would like to keep around, but can't since there isn't enough 457077b1f83SDan Magenheimer memory. So when the PFRA "evicts" a page, it first attempts to use 458140a1ef2SMichael Witten cleancache code to put the data contained in that page into 459077b1f83SDan Magenheimer "transcendent memory", memory that is not directly accessible or 460077b1f83SDan Magenheimer addressable by the kernel and is of unknown and possibly 461077b1f83SDan Magenheimer time-varying size. And when a cleancache-enabled 462077b1f83SDan Magenheimer filesystem wishes to access a page in a file on disk, it first 463077b1f83SDan Magenheimer checks cleancache to see if it already contains it; if it does, 464077b1f83SDan Magenheimer the page is copied into the kernel and a disk access is avoided. 465077b1f83SDan Magenheimer When a transcendent memory driver is available (such as zcache or 466077b1f83SDan Magenheimer Xen transcendent memory), a significant I/O reduction 467077b1f83SDan Magenheimer may be achieved. When none is available, all cleancache calls 468077b1f83SDan Magenheimer are reduced to a single pointer-compare-against-NULL resulting 469077b1f83SDan Magenheimer in a negligible performance hit. 470077b1f83SDan Magenheimer 471077b1f83SDan Magenheimer If unsure, say Y to enable cleancache 47227c6aec2SDan Magenheimer 47327c6aec2SDan Magenheimerconfig FRONTSWAP 47427c6aec2SDan Magenheimer bool "Enable frontswap to cache swap pages if tmem is present" 47527c6aec2SDan Magenheimer depends on SWAP 47627c6aec2SDan Magenheimer default n 47727c6aec2SDan Magenheimer help 47827c6aec2SDan Magenheimer Frontswap is so named because it can be thought of as the opposite 47927c6aec2SDan Magenheimer of a "backing" store for a swap device. The data is stored into 48027c6aec2SDan Magenheimer "transcendent memory", memory that is not directly accessible or 48127c6aec2SDan Magenheimer addressable by the kernel and is of unknown and possibly 48227c6aec2SDan Magenheimer time-varying size. When space in transcendent memory is available, 48327c6aec2SDan Magenheimer a significant swap I/O reduction may be achieved. When none is 48427c6aec2SDan Magenheimer available, all frontswap calls are reduced to a single pointer- 48527c6aec2SDan Magenheimer compare-against-NULL resulting in a negligible performance hit 48627c6aec2SDan Magenheimer and swap data is stored as normal on the matching swap device. 48727c6aec2SDan Magenheimer 48827c6aec2SDan Magenheimer If unsure, say Y to enable frontswap. 489f825c736SAneesh Kumar K.V 490f825c736SAneesh Kumar K.Vconfig CMA 491f825c736SAneesh Kumar K.V bool "Contiguous Memory Allocator" 492de32a817SChen Gang depends on HAVE_MEMBLOCK && MMU 493f825c736SAneesh Kumar K.V select MIGRATION 494f825c736SAneesh Kumar K.V select MEMORY_ISOLATION 495f825c736SAneesh Kumar K.V help 496f825c736SAneesh Kumar K.V This enables the Contiguous Memory Allocator which allows other 497f825c736SAneesh Kumar K.V subsystems to allocate big physically-contiguous blocks of memory. 498f825c736SAneesh Kumar K.V CMA reserves a region of memory and allows only movable pages to 499f825c736SAneesh Kumar K.V be allocated from it. This way, the kernel can use the memory for 500f825c736SAneesh Kumar K.V pagecache and when a subsystem requests for contiguous area, the 501f825c736SAneesh Kumar K.V allocated pages are migrated away to serve the contiguous request. 502f825c736SAneesh Kumar K.V 503f825c736SAneesh Kumar K.V If unsure, say "n". 504f825c736SAneesh Kumar K.V 505f825c736SAneesh Kumar K.Vconfig CMA_DEBUG 506f825c736SAneesh Kumar K.V bool "CMA debug messages (DEVELOPMENT)" 507f825c736SAneesh Kumar K.V depends on DEBUG_KERNEL && CMA 508f825c736SAneesh Kumar K.V help 509f825c736SAneesh Kumar K.V Turns on debug messages in CMA. This produces KERN_DEBUG 510f825c736SAneesh Kumar K.V messages for every CMA call as well as various messages while 511f825c736SAneesh Kumar K.V processing calls such as dma_alloc_from_contiguous(). 512f825c736SAneesh Kumar K.V This option does not affect warning and error messages. 513bf550fc9SAlexander Graf 5144e2e2770SSeth Jenningsconfig ZBUD 5154e2e2770SSeth Jennings tristate 5164e2e2770SSeth Jennings default n 5174e2e2770SSeth Jennings help 5184e2e2770SSeth Jennings A special purpose allocator for storing compressed pages. 5194e2e2770SSeth Jennings It is designed to store up to two compressed pages per physical 5204e2e2770SSeth Jennings page. While this design limits storage density, it has simple and 5214e2e2770SSeth Jennings deterministic reclaim properties that make it preferable to a higher 5224e2e2770SSeth Jennings density approach when reclaim will be used. 5234e2e2770SSeth Jennings 5242b281117SSeth Jenningsconfig ZSWAP 5252b281117SSeth Jennings bool "Compressed cache for swap pages (EXPERIMENTAL)" 5262b281117SSeth Jennings depends on FRONTSWAP && CRYPTO=y 5272b281117SSeth Jennings select CRYPTO_LZO 5282b281117SSeth Jennings select ZBUD 5292b281117SSeth Jennings default n 5302b281117SSeth Jennings help 5312b281117SSeth Jennings A lightweight compressed cache for swap pages. It takes 5322b281117SSeth Jennings pages that are in the process of being swapped out and attempts to 5332b281117SSeth Jennings compress them into a dynamically allocated RAM-based memory pool. 5342b281117SSeth Jennings This can result in a significant I/O reduction on swap device and, 5352b281117SSeth Jennings in the case where decompressing from RAM is faster that swap device 5362b281117SSeth Jennings reads, can also improve workload performance. 5372b281117SSeth Jennings 5382b281117SSeth Jennings This is marked experimental because it is a new feature (as of 5392b281117SSeth Jennings v3.11) that interacts heavily with memory reclaim. While these 5402b281117SSeth Jennings interactions don't cause any known issues on simple memory setups, 5412b281117SSeth Jennings they have not be fully explored on the large set of potential 5422b281117SSeth Jennings configurations and workloads that exist. 5432b281117SSeth Jennings 5440f8975ecSPavel Emelyanovconfig MEM_SOFT_DIRTY 5450f8975ecSPavel Emelyanov bool "Track memory changes" 546a844f386SSima Baymani depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS 5470f8975ecSPavel Emelyanov select PROC_PAGE_MONITOR 5480f8975ecSPavel Emelyanov help 5490f8975ecSPavel Emelyanov This option enables memory changes tracking by introducing a 5500f8975ecSPavel Emelyanov soft-dirty bit on pte-s. This bit it set when someone writes 5510f8975ecSPavel Emelyanov into a page just as regular dirty bit, but unlike the latter 5520f8975ecSPavel Emelyanov it can be cleared by hands. 5530f8975ecSPavel Emelyanov 5540f8975ecSPavel Emelyanov See Documentation/vm/soft-dirty.txt for more details. 555bcf1647dSMinchan Kim 556bcf1647dSMinchan Kimconfig ZSMALLOC 557bcf1647dSMinchan Kim bool "Memory allocator for compressed pages" 558bcf1647dSMinchan Kim depends on MMU 559bcf1647dSMinchan Kim default n 560bcf1647dSMinchan Kim help 561bcf1647dSMinchan Kim zsmalloc is a slab-based memory allocator designed to store 562bcf1647dSMinchan Kim compressed RAM pages. zsmalloc uses virtual memory mapping 563bcf1647dSMinchan Kim in order to reduce fragmentation. However, this results in a 564bcf1647dSMinchan Kim non-standard allocator interface where a handle, not a pointer, is 565bcf1647dSMinchan Kim returned by an alloc(). This handle must be mapped in order to 566bcf1647dSMinchan Kim access the allocated space. 567bcf1647dSMinchan Kim 568bcf1647dSMinchan Kimconfig PGTABLE_MAPPING 569bcf1647dSMinchan Kim bool "Use page table mapping to access object in zsmalloc" 570bcf1647dSMinchan Kim depends on ZSMALLOC 571bcf1647dSMinchan Kim help 572bcf1647dSMinchan Kim By default, zsmalloc uses a copy-based object mapping method to 573bcf1647dSMinchan Kim access allocations that span two pages. However, if a particular 574bcf1647dSMinchan Kim architecture (ex, ARM) performs VM mapping faster than copying, 575bcf1647dSMinchan Kim then you should select this. This causes zsmalloc to use page table 576bcf1647dSMinchan Kim mapping rather than copying for object mapping. 577bcf1647dSMinchan Kim 578bcf1647dSMinchan Kim You can check speed with zsmalloc benchmark[1]. 579bcf1647dSMinchan Kim [1] https://github.com/spartacus06/zsmalloc 580