1# SPDX-License-Identifier: GPL-2.0-only 2 3menu "Memory Management options" 4 5# 6# For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can 7# add proper SWAP support to them, in which case this can be remove. 8# 9config ARCH_NO_SWAP 10 bool 11 12config ZPOOL 13 bool 14 15menuconfig SWAP 16 bool "Support for paging of anonymous memory (swap)" 17 depends on MMU && BLOCK && !ARCH_NO_SWAP 18 default y 19 help 20 This option allows you to choose whether you want to have support 21 for so called swap devices or swap files in your kernel that are 22 used to provide more virtual memory than the actual RAM present 23 in your computer. If unsure say Y. 24 25config ZSWAP 26 bool "Compressed cache for swap pages" 27 depends on SWAP 28 select CRYPTO 29 select ZPOOL 30 help 31 A lightweight compressed cache for swap pages. It takes 32 pages that are in the process of being swapped out and attempts to 33 compress them into a dynamically allocated RAM-based memory pool. 34 This can result in a significant I/O reduction on swap device and, 35 in the case where decompressing from RAM is faster than swap device 36 reads, can also improve workload performance. 37 38config ZSWAP_DEFAULT_ON 39 bool "Enable the compressed cache for swap pages by default" 40 depends on ZSWAP 41 help 42 If selected, the compressed cache for swap pages will be enabled 43 at boot, otherwise it will be disabled. 44 45 The selection made here can be overridden by using the kernel 46 command line 'zswap.enabled=' option. 47 48config ZSWAP_SHRINKER_DEFAULT_ON 49 bool "Shrink the zswap pool on memory pressure" 50 depends on ZSWAP 51 default n 52 help 53 If selected, the zswap shrinker will be enabled, and the pages 54 stored in the zswap pool will become available for reclaim (i.e 55 written back to the backing swap device) on memory pressure. 56 57 This means that zswap writeback could happen even if the pool is 58 not yet full, or the cgroup zswap limit has not been reached, 59 reducing the chance that cold pages will reside in the zswap pool 60 and consume memory indefinitely. 61 62choice 63 prompt "Default compressor" 64 depends on ZSWAP 65 default ZSWAP_COMPRESSOR_DEFAULT_LZO 66 help 67 Selects the default compression algorithm for the compressed cache 68 for swap pages. 69 70 For an overview what kind of performance can be expected from 71 a particular compression algorithm please refer to the benchmarks 72 available at the following LWN page: 73 https://lwn.net/Articles/751795/ 74 75 If in doubt, select 'LZO'. 76 77 The selection made here can be overridden by using the kernel 78 command line 'zswap.compressor=' option. 79 80config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE 81 bool "Deflate" 82 select CRYPTO_DEFLATE 83 help 84 Use the Deflate algorithm as the default compression algorithm. 85 86config ZSWAP_COMPRESSOR_DEFAULT_LZO 87 bool "LZO" 88 select CRYPTO_LZO 89 help 90 Use the LZO algorithm as the default compression algorithm. 91 92config ZSWAP_COMPRESSOR_DEFAULT_842 93 bool "842" 94 select CRYPTO_842 95 help 96 Use the 842 algorithm as the default compression algorithm. 97 98config ZSWAP_COMPRESSOR_DEFAULT_LZ4 99 bool "LZ4" 100 select CRYPTO_LZ4 101 help 102 Use the LZ4 algorithm as the default compression algorithm. 103 104config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC 105 bool "LZ4HC" 106 select CRYPTO_LZ4HC 107 help 108 Use the LZ4HC algorithm as the default compression algorithm. 109 110config ZSWAP_COMPRESSOR_DEFAULT_ZSTD 111 bool "zstd" 112 select CRYPTO_ZSTD 113 help 114 Use the zstd algorithm as the default compression algorithm. 115endchoice 116 117config ZSWAP_COMPRESSOR_DEFAULT 118 string 119 depends on ZSWAP 120 default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE 121 default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO 122 default "842" if ZSWAP_COMPRESSOR_DEFAULT_842 123 default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4 124 default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC 125 default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD 126 default "" 127 128choice 129 prompt "Default allocator" 130 depends on ZSWAP 131 default ZSWAP_ZPOOL_DEFAULT_ZSMALLOC if MMU 132 default ZSWAP_ZPOOL_DEFAULT_ZBUD 133 help 134 Selects the default allocator for the compressed cache for 135 swap pages. 136 The default is 'zbud' for compatibility, however please do 137 read the description of each of the allocators below before 138 making a right choice. 139 140 The selection made here can be overridden by using the kernel 141 command line 'zswap.zpool=' option. 142 143config ZSWAP_ZPOOL_DEFAULT_ZBUD 144 bool "zbud" 145 select ZBUD 146 help 147 Use the zbud allocator as the default allocator. 148 149config ZSWAP_ZPOOL_DEFAULT_Z3FOLD_DEPRECATED 150 bool "z3foldi (DEPRECATED)" 151 select Z3FOLD_DEPRECATED 152 help 153 Use the z3fold allocator as the default allocator. 154 155 Deprecated and scheduled for removal in a few cycles, 156 see CONFIG_Z3FOLD_DEPRECATED. 157 158config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC 159 bool "zsmalloc" 160 select ZSMALLOC 161 help 162 Use the zsmalloc allocator as the default allocator. 163endchoice 164 165config ZSWAP_ZPOOL_DEFAULT 166 string 167 depends on ZSWAP 168 default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD 169 default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD_DEPRECATED 170 default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC 171 default "" 172 173config ZBUD 174 tristate "2:1 compression allocator (zbud)" 175 depends on ZSWAP 176 help 177 A special purpose allocator for storing compressed pages. 178 It is designed to store up to two compressed pages per physical 179 page. While this design limits storage density, it has simple and 180 deterministic reclaim properties that make it preferable to a higher 181 density approach when reclaim will be used. 182 183config Z3FOLD_DEPRECATED 184 tristate "3:1 compression allocator (z3fold) (DEPRECATED)" 185 depends on ZSWAP 186 help 187 Deprecated and scheduled for removal in a few cycles. If you have 188 a good reason for using Z3FOLD over ZSMALLOC, please contact 189 linux-mm@kvack.org and the zswap maintainers. 190 191 A special purpose allocator for storing compressed pages. 192 It is designed to store up to three compressed pages per physical 193 page. It is a ZBUD derivative so the simplicity and determinism are 194 still there. 195 196config Z3FOLD 197 tristate 198 default y if Z3FOLD_DEPRECATED=y 199 default m if Z3FOLD_DEPRECATED=m 200 depends on Z3FOLD_DEPRECATED 201 202config ZSMALLOC 203 tristate 204 prompt "N:1 compression allocator (zsmalloc)" if (ZSWAP || ZRAM) 205 depends on MMU 206 help 207 zsmalloc is a slab-based memory allocator designed to store 208 pages of various compression levels efficiently. It achieves 209 the highest storage density with the least amount of fragmentation. 210 211config ZSMALLOC_STAT 212 bool "Export zsmalloc statistics" 213 depends on ZSMALLOC 214 select DEBUG_FS 215 help 216 This option enables code in the zsmalloc to collect various 217 statistics about what's happening in zsmalloc and exports that 218 information to userspace via debugfs. 219 If unsure, say N. 220 221config ZSMALLOC_CHAIN_SIZE 222 int "Maximum number of physical pages per-zspage" 223 default 8 224 range 4 16 225 depends on ZSMALLOC 226 help 227 This option sets the upper limit on the number of physical pages 228 that a zmalloc page (zspage) can consist of. The optimal zspage 229 chain size is calculated for each size class during the 230 initialization of the pool. 231 232 Changing this option can alter the characteristics of size classes, 233 such as the number of pages per zspage and the number of objects 234 per zspage. This can also result in different configurations of 235 the pool, as zsmalloc merges size classes with similar 236 characteristics. 237 238 For more information, see zsmalloc documentation. 239 240menu "Slab allocator options" 241 242config SLUB 243 def_bool y 244 245config SLUB_TINY 246 bool "Configure for minimal memory footprint" 247 depends on EXPERT 248 select SLAB_MERGE_DEFAULT 249 help 250 Configures the slab allocator in a way to achieve minimal memory 251 footprint, sacrificing scalability, debugging and other features. 252 This is intended only for the smallest system that had used the 253 SLOB allocator and is not recommended for systems with more than 254 16MB RAM. 255 256 If unsure, say N. 257 258config SLAB_MERGE_DEFAULT 259 bool "Allow slab caches to be merged" 260 default y 261 help 262 For reduced kernel memory fragmentation, slab caches can be 263 merged when they share the same size and other characteristics. 264 This carries a risk of kernel heap overflows being able to 265 overwrite objects from merged caches (and more easily control 266 cache layout), which makes such heap attacks easier to exploit 267 by attackers. By keeping caches unmerged, these kinds of exploits 268 can usually only damage objects in the same cache. To disable 269 merging at runtime, "slab_nomerge" can be passed on the kernel 270 command line. 271 272config SLAB_FREELIST_RANDOM 273 bool "Randomize slab freelist" 274 depends on !SLUB_TINY 275 help 276 Randomizes the freelist order used on creating new pages. This 277 security feature reduces the predictability of the kernel slab 278 allocator against heap overflows. 279 280config SLAB_FREELIST_HARDENED 281 bool "Harden slab freelist metadata" 282 depends on !SLUB_TINY 283 help 284 Many kernel heap attacks try to target slab cache metadata and 285 other infrastructure. This options makes minor performance 286 sacrifices to harden the kernel slab allocator against common 287 freelist exploit methods. 288 289config SLAB_BUCKETS 290 bool "Support allocation from separate kmalloc buckets" 291 depends on !SLUB_TINY 292 default SLAB_FREELIST_HARDENED 293 help 294 Kernel heap attacks frequently depend on being able to create 295 specifically-sized allocations with user-controlled contents 296 that will be allocated into the same kmalloc bucket as a 297 target object. To avoid sharing these allocation buckets, 298 provide an explicitly separated set of buckets to be used for 299 user-controlled allocations. This may very slightly increase 300 memory fragmentation, though in practice it's only a handful 301 of extra pages since the bulk of user-controlled allocations 302 are relatively long-lived. 303 304 If unsure, say Y. 305 306config SLUB_STATS 307 default n 308 bool "Enable performance statistics" 309 depends on SYSFS && !SLUB_TINY 310 help 311 The statistics are useful to debug slab allocation behavior in 312 order find ways to optimize the allocator. This should never be 313 enabled for production use since keeping statistics slows down 314 the allocator by a few percentage points. The slabinfo command 315 supports the determination of the most active slabs to figure 316 out which slabs are relevant to a particular load. 317 Try running: slabinfo -DA 318 319config SLUB_CPU_PARTIAL 320 default y 321 depends on SMP && !SLUB_TINY 322 bool "Enable per cpu partial caches" 323 help 324 Per cpu partial caches accelerate objects allocation and freeing 325 that is local to a processor at the price of more indeterminism 326 in the latency of the free. On overflow these caches will be cleared 327 which requires the taking of locks that may cause latency spikes. 328 Typically one would choose no for a realtime system. 329 330config RANDOM_KMALLOC_CACHES 331 default n 332 depends on !SLUB_TINY 333 bool "Randomize slab caches for normal kmalloc" 334 help 335 A hardening feature that creates multiple copies of slab caches for 336 normal kmalloc allocation and makes kmalloc randomly pick one based 337 on code address, which makes the attackers more difficult to spray 338 vulnerable memory objects on the heap for the purpose of exploiting 339 memory vulnerabilities. 340 341 Currently the number of copies is set to 16, a reasonably large value 342 that effectively diverges the memory objects allocated for different 343 subsystems or modules into different caches, at the expense of a 344 limited degree of memory and CPU overhead that relates to hardware and 345 system workload. 346 347endmenu # Slab allocator options 348 349config SHUFFLE_PAGE_ALLOCATOR 350 bool "Page allocator randomization" 351 default SLAB_FREELIST_RANDOM && ACPI_NUMA 352 help 353 Randomization of the page allocator improves the average 354 utilization of a direct-mapped memory-side-cache. See section 355 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI 356 6.2a specification for an example of how a platform advertises 357 the presence of a memory-side-cache. There are also incidental 358 security benefits as it reduces the predictability of page 359 allocations to compliment SLAB_FREELIST_RANDOM, but the 360 default granularity of shuffling on the MAX_PAGE_ORDER i.e, 10th 361 order of pages is selected based on cache utilization benefits 362 on x86. 363 364 While the randomization improves cache utilization it may 365 negatively impact workloads on platforms without a cache. For 366 this reason, by default, the randomization is not enabled even 367 if SHUFFLE_PAGE_ALLOCATOR=y. The randomization may be force enabled 368 with the 'page_alloc.shuffle' kernel command line parameter. 369 370 Say Y if unsure. 371 372config COMPAT_BRK 373 bool "Disable heap randomization" 374 default y 375 help 376 Randomizing heap placement makes heap exploits harder, but it 377 also breaks ancient binaries (including anything libc5 based). 378 This option changes the bootup default to heap randomization 379 disabled, and can be overridden at runtime by setting 380 /proc/sys/kernel/randomize_va_space to 2. 381 382 On non-ancient distros (post-2000 ones) N is usually a safe choice. 383 384config MMAP_ALLOW_UNINITIALIZED 385 bool "Allow mmapped anonymous memory to be uninitialized" 386 depends on EXPERT && !MMU 387 default n 388 help 389 Normally, and according to the Linux spec, anonymous memory obtained 390 from mmap() has its contents cleared before it is passed to 391 userspace. Enabling this config option allows you to request that 392 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus 393 providing a huge performance boost. If this option is not enabled, 394 then the flag will be ignored. 395 396 This is taken advantage of by uClibc's malloc(), and also by 397 ELF-FDPIC binfmt's brk and stack allocator. 398 399 Because of the obvious security issues, this option should only be 400 enabled on embedded devices where you control what is run in 401 userspace. Since that isn't generally a problem on no-MMU systems, 402 it is normally safe to say Y here. 403 404 See Documentation/admin-guide/mm/nommu-mmap.rst for more information. 405 406config SELECT_MEMORY_MODEL 407 def_bool y 408 depends on ARCH_SELECT_MEMORY_MODEL 409 410choice 411 prompt "Memory model" 412 depends on SELECT_MEMORY_MODEL 413 default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT 414 default FLATMEM_MANUAL 415 help 416 This option allows you to change some of the ways that 417 Linux manages its memory internally. Most users will 418 only have one option here selected by the architecture 419 configuration. This is normal. 420 421config FLATMEM_MANUAL 422 bool "Flat Memory" 423 depends on !ARCH_SPARSEMEM_ENABLE || ARCH_FLATMEM_ENABLE 424 help 425 This option is best suited for non-NUMA systems with 426 flat address space. The FLATMEM is the most efficient 427 system in terms of performance and resource consumption 428 and it is the best option for smaller systems. 429 430 For systems that have holes in their physical address 431 spaces and for features like NUMA and memory hotplug, 432 choose "Sparse Memory". 433 434 If unsure, choose this option (Flat Memory) over any other. 435 436config SPARSEMEM_MANUAL 437 bool "Sparse Memory" 438 depends on ARCH_SPARSEMEM_ENABLE 439 help 440 This will be the only option for some systems, including 441 memory hot-plug systems. This is normal. 442 443 This option provides efficient support for systems with 444 holes is their physical address space and allows memory 445 hot-plug and hot-remove. 446 447 If unsure, choose "Flat Memory" over this option. 448 449endchoice 450 451config SPARSEMEM 452 def_bool y 453 depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL 454 455config FLATMEM 456 def_bool y 457 depends on !SPARSEMEM || FLATMEM_MANUAL 458 459# 460# SPARSEMEM_EXTREME (which is the default) does some bootmem 461# allocations when sparse_init() is called. If this cannot 462# be done on your architecture, select this option. However, 463# statically allocating the mem_section[] array can potentially 464# consume vast quantities of .bss, so be careful. 465# 466# This option will also potentially produce smaller runtime code 467# with gcc 3.4 and later. 468# 469config SPARSEMEM_STATIC 470 bool 471 472# 473# Architecture platforms which require a two level mem_section in SPARSEMEM 474# must select this option. This is usually for architecture platforms with 475# an extremely sparse physical address space. 476# 477config SPARSEMEM_EXTREME 478 def_bool y 479 depends on SPARSEMEM && !SPARSEMEM_STATIC 480 481config SPARSEMEM_VMEMMAP_ENABLE 482 bool 483 484config SPARSEMEM_VMEMMAP 485 bool "Sparse Memory virtual memmap" 486 depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE 487 default y 488 help 489 SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise 490 pfn_to_page and page_to_pfn operations. This is the most 491 efficient option when sufficient kernel resources are available. 492# 493# Select this config option from the architecture Kconfig, if it is preferred 494# to enable the feature of HugeTLB/dev_dax vmemmap optimization. 495# 496config ARCH_WANT_OPTIMIZE_DAX_VMEMMAP 497 bool 498 499config ARCH_WANT_OPTIMIZE_HUGETLB_VMEMMAP 500 bool 501 502config HAVE_MEMBLOCK_PHYS_MAP 503 bool 504 505config HAVE_GUP_FAST 506 depends on MMU 507 bool 508 509# Don't discard allocated memory used to track "memory" and "reserved" memblocks 510# after early boot, so it can still be used to test for validity of memory. 511# Also, memblocks are updated with memory hot(un)plug. 512config ARCH_KEEP_MEMBLOCK 513 bool 514 515# Keep arch NUMA mapping infrastructure post-init. 516config NUMA_KEEP_MEMINFO 517 bool 518 519config MEMORY_ISOLATION 520 bool 521 522# IORESOURCE_SYSTEM_RAM regions in the kernel resource tree that are marked 523# IORESOURCE_EXCLUSIVE cannot be mapped to user space, for example, via 524# /dev/mem. 525config EXCLUSIVE_SYSTEM_RAM 526 def_bool y 527 depends on !DEVMEM || STRICT_DEVMEM 528 529# 530# Only be set on architectures that have completely implemented memory hotplug 531# feature. If you are not sure, don't touch it. 532# 533config HAVE_BOOTMEM_INFO_NODE 534 def_bool n 535 536config ARCH_ENABLE_MEMORY_HOTPLUG 537 bool 538 539config ARCH_ENABLE_MEMORY_HOTREMOVE 540 bool 541 542# eventually, we can have this option just 'select SPARSEMEM' 543menuconfig MEMORY_HOTPLUG 544 bool "Memory hotplug" 545 select MEMORY_ISOLATION 546 depends on SPARSEMEM 547 depends on ARCH_ENABLE_MEMORY_HOTPLUG 548 depends on 64BIT 549 select NUMA_KEEP_MEMINFO if NUMA 550 551if MEMORY_HOTPLUG 552 553config MEMORY_HOTPLUG_DEFAULT_ONLINE 554 bool "Online the newly added memory blocks by default" 555 depends on MEMORY_HOTPLUG 556 help 557 This option sets the default policy setting for memory hotplug 558 onlining policy (/sys/devices/system/memory/auto_online_blocks) which 559 determines what happens to newly added memory regions. Policy setting 560 can always be changed at runtime. 561 See Documentation/admin-guide/mm/memory-hotplug.rst for more information. 562 563 Say Y here if you want all hot-plugged memory blocks to appear in 564 'online' state by default. 565 Say N here if you want the default policy to keep all hot-plugged 566 memory blocks in 'offline' state. 567 568config MEMORY_HOTREMOVE 569 bool "Allow for memory hot remove" 570 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64) 571 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE 572 depends on MIGRATION 573 574config MHP_MEMMAP_ON_MEMORY 575 def_bool y 576 depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP 577 depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE 578 579endif # MEMORY_HOTPLUG 580 581config ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE 582 bool 583 584# Heavily threaded applications may benefit from splitting the mm-wide 585# page_table_lock, so that faults on different parts of the user address 586# space can be handled with less contention: split it at this NR_CPUS. 587# Default to 4 for wider testing, though 8 might be more appropriate. 588# ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock. 589# PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes. 590# SPARC32 allocates multiple pte tables within a single page, and therefore 591# a per-page lock leads to problems when multiple tables need to be locked 592# at the same time (e.g. copy_page_range()). 593# DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page. 594# 595config SPLIT_PTE_PTLOCKS 596 def_bool y 597 depends on MMU 598 depends on NR_CPUS >= 4 599 depends on !ARM || CPU_CACHE_VIPT 600 depends on !PARISC || PA20 601 depends on !SPARC32 602 603config ARCH_ENABLE_SPLIT_PMD_PTLOCK 604 bool 605 606config SPLIT_PMD_PTLOCKS 607 def_bool y 608 depends on SPLIT_PTE_PTLOCKS && ARCH_ENABLE_SPLIT_PMD_PTLOCK 609 610# 611# support for memory balloon 612config MEMORY_BALLOON 613 bool 614 615# 616# support for memory balloon compaction 617config BALLOON_COMPACTION 618 bool "Allow for balloon memory compaction/migration" 619 default y 620 depends on COMPACTION && MEMORY_BALLOON 621 help 622 Memory fragmentation introduced by ballooning might reduce 623 significantly the number of 2MB contiguous memory blocks that can be 624 used within a guest, thus imposing performance penalties associated 625 with the reduced number of transparent huge pages that could be used 626 by the guest workload. Allowing the compaction & migration for memory 627 pages enlisted as being part of memory balloon devices avoids the 628 scenario aforementioned and helps improving memory defragmentation. 629 630# 631# support for memory compaction 632config COMPACTION 633 bool "Allow for memory compaction" 634 default y 635 select MIGRATION 636 depends on MMU 637 help 638 Compaction is the only memory management component to form 639 high order (larger physically contiguous) memory blocks 640 reliably. The page allocator relies on compaction heavily and 641 the lack of the feature can lead to unexpected OOM killer 642 invocations for high order memory requests. You shouldn't 643 disable this option unless there really is a strong reason for 644 it and then we would be really interested to hear about that at 645 linux-mm@kvack.org. 646 647config COMPACT_UNEVICTABLE_DEFAULT 648 int 649 depends on COMPACTION 650 default 0 if PREEMPT_RT 651 default 1 652 653# 654# support for free page reporting 655config PAGE_REPORTING 656 bool "Free page reporting" 657 help 658 Free page reporting allows for the incremental acquisition of 659 free pages from the buddy allocator for the purpose of reporting 660 those pages to another entity, such as a hypervisor, so that the 661 memory can be freed within the host for other uses. 662 663# 664# support for page migration 665# 666config MIGRATION 667 bool "Page migration" 668 default y 669 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU 670 help 671 Allows the migration of the physical location of pages of processes 672 while the virtual addresses are not changed. This is useful in 673 two situations. The first is on NUMA systems to put pages nearer 674 to the processors accessing. The second is when allocating huge 675 pages as migration can relocate pages to satisfy a huge page 676 allocation instead of reclaiming. 677 678config DEVICE_MIGRATION 679 def_bool MIGRATION && ZONE_DEVICE 680 681config ARCH_ENABLE_HUGEPAGE_MIGRATION 682 bool 683 684config ARCH_ENABLE_THP_MIGRATION 685 bool 686 687config HUGETLB_PAGE_SIZE_VARIABLE 688 def_bool n 689 help 690 Allows the pageblock_order value to be dynamic instead of just standard 691 HUGETLB_PAGE_ORDER when there are multiple HugeTLB page sizes available 692 on a platform. 693 694 Note that the pageblock_order cannot exceed MAX_PAGE_ORDER and will be 695 clamped down to MAX_PAGE_ORDER. 696 697config CONTIG_ALLOC 698 def_bool (MEMORY_ISOLATION && COMPACTION) || CMA 699 700config PCP_BATCH_SCALE_MAX 701 int "Maximum scale factor of PCP (Per-CPU pageset) batch allocate/free" 702 default 5 703 range 0 6 704 help 705 In page allocator, PCP (Per-CPU pageset) is refilled and drained in 706 batches. The batch number is scaled automatically to improve page 707 allocation/free throughput. But too large scale factor may hurt 708 latency. This option sets the upper limit of scale factor to limit 709 the maximum latency. 710 711config PHYS_ADDR_T_64BIT 712 def_bool 64BIT 713 714config BOUNCE 715 bool "Enable bounce buffers" 716 default y 717 depends on BLOCK && MMU && HIGHMEM 718 help 719 Enable bounce buffers for devices that cannot access the full range of 720 memory available to the CPU. Enabled by default when HIGHMEM is 721 selected, but you may say n to override this. 722 723config MMU_NOTIFIER 724 bool 725 select INTERVAL_TREE 726 727config KSM 728 bool "Enable KSM for page merging" 729 depends on MMU 730 select XXHASH 731 help 732 Enable Kernel Samepage Merging: KSM periodically scans those areas 733 of an application's address space that an app has advised may be 734 mergeable. When it finds pages of identical content, it replaces 735 the many instances by a single page with that content, so 736 saving memory until one or another app needs to modify the content. 737 Recommended for use with KVM, or with other duplicative applications. 738 See Documentation/mm/ksm.rst for more information: KSM is inactive 739 until a program has madvised that an area is MADV_MERGEABLE, and 740 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). 741 742config DEFAULT_MMAP_MIN_ADDR 743 int "Low address space to protect from user allocation" 744 depends on MMU 745 default 4096 746 help 747 This is the portion of low virtual memory which should be protected 748 from userspace allocation. Keeping a user from writing to low pages 749 can help reduce the impact of kernel NULL pointer bugs. 750 751 For most arm64, ppc64 and x86 users with lots of address space 752 a value of 65536 is reasonable and should cause no problems. 753 On arm and other archs it should not be higher than 32768. 754 Programs which use vm86 functionality or have some need to map 755 this low address space will need CAP_SYS_RAWIO or disable this 756 protection by setting the value to 0. 757 758 This value can be changed after boot using the 759 /proc/sys/vm/mmap_min_addr tunable. 760 761config ARCH_SUPPORTS_MEMORY_FAILURE 762 bool 763 764config MEMORY_FAILURE 765 depends on MMU 766 depends on ARCH_SUPPORTS_MEMORY_FAILURE 767 bool "Enable recovery from hardware memory errors" 768 select MEMORY_ISOLATION 769 select RAS 770 help 771 Enables code to recover from some memory failures on systems 772 with MCA recovery. This allows a system to continue running 773 even when some of its memory has uncorrected errors. This requires 774 special hardware support and typically ECC memory. 775 776config HWPOISON_INJECT 777 tristate "HWPoison pages injector" 778 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS 779 select PROC_PAGE_MONITOR 780 781config NOMMU_INITIAL_TRIM_EXCESS 782 int "Turn on mmap() excess space trimming before booting" 783 depends on !MMU 784 default 1 785 help 786 The NOMMU mmap() frequently needs to allocate large contiguous chunks 787 of memory on which to store mappings, but it can only ask the system 788 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently 789 more than it requires. To deal with this, mmap() is able to trim off 790 the excess and return it to the allocator. 791 792 If trimming is enabled, the excess is trimmed off and returned to the 793 system allocator, which can cause extra fragmentation, particularly 794 if there are a lot of transient processes. 795 796 If trimming is disabled, the excess is kept, but not used, which for 797 long-term mappings means that the space is wasted. 798 799 Trimming can be dynamically controlled through a sysctl option 800 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of 801 excess pages there must be before trimming should occur, or zero if 802 no trimming is to occur. 803 804 This option specifies the initial value of this option. The default 805 of 1 says that all excess pages should be trimmed. 806 807 See Documentation/admin-guide/mm/nommu-mmap.rst for more information. 808 809config ARCH_WANT_GENERAL_HUGETLB 810 bool 811 812config ARCH_WANTS_THP_SWAP 813 def_bool n 814 815menuconfig TRANSPARENT_HUGEPAGE 816 bool "Transparent Hugepage Support" 817 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE && !PREEMPT_RT 818 select COMPACTION 819 select XARRAY_MULTI 820 help 821 Transparent Hugepages allows the kernel to use huge pages and 822 huge tlb transparently to the applications whenever possible. 823 This feature can improve computing performance to certain 824 applications by speeding up page faults during memory 825 allocation, by reducing the number of tlb misses and by speeding 826 up the pagetable walking. 827 828 If memory constrained on embedded, you may want to say N. 829 830if TRANSPARENT_HUGEPAGE 831 832choice 833 prompt "Transparent Hugepage Support sysfs defaults" 834 depends on TRANSPARENT_HUGEPAGE 835 default TRANSPARENT_HUGEPAGE_ALWAYS 836 help 837 Selects the sysfs defaults for Transparent Hugepage Support. 838 839 config TRANSPARENT_HUGEPAGE_ALWAYS 840 bool "always" 841 help 842 Enabling Transparent Hugepage always, can increase the 843 memory footprint of applications without a guaranteed 844 benefit but it will work automatically for all applications. 845 846 config TRANSPARENT_HUGEPAGE_MADVISE 847 bool "madvise" 848 help 849 Enabling Transparent Hugepage madvise, will only provide a 850 performance improvement benefit to the applications using 851 madvise(MADV_HUGEPAGE) but it won't risk to increase the 852 memory footprint of applications without a guaranteed 853 benefit. 854 855 config TRANSPARENT_HUGEPAGE_NEVER 856 bool "never" 857 help 858 Disable Transparent Hugepage by default. It can still be 859 enabled at runtime via sysfs. 860endchoice 861 862config THP_SWAP 863 def_bool y 864 depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP && 64BIT 865 help 866 Swap transparent huge pages in one piece, without splitting. 867 XXX: For now, swap cluster backing transparent huge page 868 will be split after swapout. 869 870 For selection by architectures with reasonable THP sizes. 871 872config READ_ONLY_THP_FOR_FS 873 bool "Read-only THP for filesystems (EXPERIMENTAL)" 874 depends on TRANSPARENT_HUGEPAGE && SHMEM 875 876 help 877 Allow khugepaged to put read-only file-backed pages in THP. 878 879 This is marked experimental because it is a new feature. Write 880 support of file THPs will be developed in the next few release 881 cycles. 882 883endif # TRANSPARENT_HUGEPAGE 884 885# 886# The architecture supports pgtable leaves that is larger than PAGE_SIZE 887# 888config PGTABLE_HAS_HUGE_LEAVES 889 def_bool TRANSPARENT_HUGEPAGE || HUGETLB_PAGE 890 891# TODO: Allow to be enabled without THP 892config ARCH_SUPPORTS_HUGE_PFNMAP 893 def_bool n 894 depends on TRANSPARENT_HUGEPAGE 895 896config ARCH_SUPPORTS_PMD_PFNMAP 897 def_bool y 898 depends on ARCH_SUPPORTS_HUGE_PFNMAP && HAVE_ARCH_TRANSPARENT_HUGEPAGE 899 900config ARCH_SUPPORTS_PUD_PFNMAP 901 def_bool y 902 depends on ARCH_SUPPORTS_HUGE_PFNMAP && HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD 903 904# 905# UP and nommu archs use km based percpu allocator 906# 907config NEED_PER_CPU_KM 908 depends on !SMP || !MMU 909 bool 910 default y 911 912config NEED_PER_CPU_EMBED_FIRST_CHUNK 913 bool 914 915config NEED_PER_CPU_PAGE_FIRST_CHUNK 916 bool 917 918config USE_PERCPU_NUMA_NODE_ID 919 bool 920 921config HAVE_SETUP_PER_CPU_AREA 922 bool 923 924config CMA 925 bool "Contiguous Memory Allocator" 926 depends on MMU 927 select MIGRATION 928 select MEMORY_ISOLATION 929 help 930 This enables the Contiguous Memory Allocator which allows other 931 subsystems to allocate big physically-contiguous blocks of memory. 932 CMA reserves a region of memory and allows only movable pages to 933 be allocated from it. This way, the kernel can use the memory for 934 pagecache and when a subsystem requests for contiguous area, the 935 allocated pages are migrated away to serve the contiguous request. 936 937 If unsure, say "n". 938 939config CMA_DEBUGFS 940 bool "CMA debugfs interface" 941 depends on CMA && DEBUG_FS 942 help 943 Turns on the DebugFS interface for CMA. 944 945config CMA_SYSFS 946 bool "CMA information through sysfs interface" 947 depends on CMA && SYSFS 948 help 949 This option exposes some sysfs attributes to get information 950 from CMA. 951 952config CMA_AREAS 953 int "Maximum count of the CMA areas" 954 depends on CMA 955 default 20 if NUMA 956 default 8 957 help 958 CMA allows to create CMA areas for particular purpose, mainly, 959 used as device private area. This parameter sets the maximum 960 number of CMA area in the system. 961 962 If unsure, leave the default value "8" in UMA and "20" in NUMA. 963 964config MEM_SOFT_DIRTY 965 bool "Track memory changes" 966 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS 967 select PROC_PAGE_MONITOR 968 help 969 This option enables memory changes tracking by introducing a 970 soft-dirty bit on pte-s. This bit it set when someone writes 971 into a page just as regular dirty bit, but unlike the latter 972 it can be cleared by hands. 973 974 See Documentation/admin-guide/mm/soft-dirty.rst for more details. 975 976config GENERIC_EARLY_IOREMAP 977 bool 978 979config STACK_MAX_DEFAULT_SIZE_MB 980 int "Default maximum user stack size for 32-bit processes (MB)" 981 default 100 982 range 8 2048 983 depends on STACK_GROWSUP && (!64BIT || COMPAT) 984 help 985 This is the maximum stack size in Megabytes in the VM layout of 32-bit 986 user processes when the stack grows upwards (currently only on parisc 987 arch) when the RLIMIT_STACK hard limit is unlimited. 988 989 A sane initial value is 100 MB. 990 991config DEFERRED_STRUCT_PAGE_INIT 992 bool "Defer initialisation of struct pages to kthreads" 993 depends on SPARSEMEM 994 depends on !NEED_PER_CPU_KM 995 depends on 64BIT 996 depends on !KMSAN 997 select PADATA 998 help 999 Ordinarily all struct pages are initialised during early boot in a 1000 single thread. On very large machines this can take a considerable 1001 amount of time. If this option is set, large machines will bring up 1002 a subset of memmap at boot and then initialise the rest in parallel. 1003 This has a potential performance impact on tasks running early in the 1004 lifetime of the system until these kthreads finish the 1005 initialisation. 1006 1007config PAGE_IDLE_FLAG 1008 bool 1009 select PAGE_EXTENSION if !64BIT 1010 help 1011 This adds PG_idle and PG_young flags to 'struct page'. PTE Accessed 1012 bit writers can set the state of the bit in the flags so that PTE 1013 Accessed bit readers may avoid disturbance. 1014 1015config IDLE_PAGE_TRACKING 1016 bool "Enable idle page tracking" 1017 depends on SYSFS && MMU 1018 select PAGE_IDLE_FLAG 1019 help 1020 This feature allows to estimate the amount of user pages that have 1021 not been touched during a given period of time. This information can 1022 be useful to tune memory cgroup limits and/or for job placement 1023 within a compute cluster. 1024 1025 See Documentation/admin-guide/mm/idle_page_tracking.rst for 1026 more details. 1027 1028# Architectures which implement cpu_dcache_is_aliasing() to query 1029# whether the data caches are aliased (VIVT or VIPT with dcache 1030# aliasing) need to select this. 1031config ARCH_HAS_CPU_CACHE_ALIASING 1032 bool 1033 1034config ARCH_HAS_CACHE_LINE_SIZE 1035 bool 1036 1037config ARCH_HAS_CURRENT_STACK_POINTER 1038 bool 1039 help 1040 In support of HARDENED_USERCOPY performing stack variable lifetime 1041 checking, an architecture-agnostic way to find the stack pointer 1042 is needed. Once an architecture defines an unsigned long global 1043 register alias named "current_stack_pointer", this config can be 1044 selected. 1045 1046config ARCH_HAS_PTE_DEVMAP 1047 bool 1048 1049config ARCH_HAS_ZONE_DMA_SET 1050 bool 1051 1052config ZONE_DMA 1053 bool "Support DMA zone" if ARCH_HAS_ZONE_DMA_SET 1054 default y if ARM64 || X86 1055 1056config ZONE_DMA32 1057 bool "Support DMA32 zone" if ARCH_HAS_ZONE_DMA_SET 1058 depends on !X86_32 1059 default y if ARM64 1060 1061config ZONE_DEVICE 1062 bool "Device memory (pmem, HMM, etc...) hotplug support" 1063 depends on MEMORY_HOTPLUG 1064 depends on MEMORY_HOTREMOVE 1065 depends on SPARSEMEM_VMEMMAP 1066 depends on ARCH_HAS_PTE_DEVMAP 1067 select XARRAY_MULTI 1068 1069 help 1070 Device memory hotplug support allows for establishing pmem, 1071 or other device driver discovered memory regions, in the 1072 memmap. This allows pfn_to_page() lookups of otherwise 1073 "device-physical" addresses which is needed for using a DAX 1074 mapping in an O_DIRECT operation, among other things. 1075 1076 If FS_DAX is enabled, then say Y. 1077 1078# 1079# Helpers to mirror range of the CPU page tables of a process into device page 1080# tables. 1081# 1082config HMM_MIRROR 1083 bool 1084 depends on MMU 1085 1086config GET_FREE_REGION 1087 depends on SPARSEMEM 1088 bool 1089 1090config DEVICE_PRIVATE 1091 bool "Unaddressable device memory (GPU memory, ...)" 1092 depends on ZONE_DEVICE 1093 select GET_FREE_REGION 1094 1095 help 1096 Allows creation of struct pages to represent unaddressable device 1097 memory; i.e., memory that is only accessible from the device (or 1098 group of devices). You likely also want to select HMM_MIRROR. 1099 1100config VMAP_PFN 1101 bool 1102 1103config ARCH_USES_HIGH_VMA_FLAGS 1104 bool 1105config ARCH_HAS_PKEYS 1106 bool 1107 1108config ARCH_USES_PG_ARCH_2 1109 bool 1110config ARCH_USES_PG_ARCH_3 1111 bool 1112 1113config VM_EVENT_COUNTERS 1114 default y 1115 bool "Enable VM event counters for /proc/vmstat" if EXPERT 1116 help 1117 VM event counters are needed for event counts to be shown. 1118 This option allows the disabling of the VM event counters 1119 on EXPERT systems. /proc/vmstat will only show page counts 1120 if VM event counters are disabled. 1121 1122config PERCPU_STATS 1123 bool "Collect percpu memory statistics" 1124 help 1125 This feature collects and exposes statistics via debugfs. The 1126 information includes global and per chunk statistics, which can 1127 be used to help understand percpu memory usage. 1128 1129config GUP_TEST 1130 bool "Enable infrastructure for get_user_pages()-related unit tests" 1131 depends on DEBUG_FS 1132 help 1133 Provides /sys/kernel/debug/gup_test, which in turn provides a way 1134 to make ioctl calls that can launch kernel-based unit tests for 1135 the get_user_pages*() and pin_user_pages*() family of API calls. 1136 1137 These tests include benchmark testing of the _fast variants of 1138 get_user_pages*() and pin_user_pages*(), as well as smoke tests of 1139 the non-_fast variants. 1140 1141 There is also a sub-test that allows running dump_page() on any 1142 of up to eight pages (selected by command line args) within the 1143 range of user-space addresses. These pages are either pinned via 1144 pin_user_pages*(), or pinned via get_user_pages*(), as specified 1145 by other command line arguments. 1146 1147 See tools/testing/selftests/mm/gup_test.c 1148 1149comment "GUP_TEST needs to have DEBUG_FS enabled" 1150 depends on !GUP_TEST && !DEBUG_FS 1151 1152config GUP_GET_PXX_LOW_HIGH 1153 bool 1154 1155config DMAPOOL_TEST 1156 tristate "Enable a module to run time tests on dma_pool" 1157 depends on HAS_DMA 1158 help 1159 Provides a test module that will allocate and free many blocks of 1160 various sizes and report how long it takes. This is intended to 1161 provide a consistent way to measure how changes to the 1162 dma_pool_alloc/free routines affect performance. 1163 1164config ARCH_HAS_PTE_SPECIAL 1165 bool 1166 1167config MAPPING_DIRTY_HELPERS 1168 bool 1169 1170config KMAP_LOCAL 1171 bool 1172 1173config KMAP_LOCAL_NON_LINEAR_PTE_ARRAY 1174 bool 1175 1176# struct io_mapping based helper. Selected by drivers that need them 1177config IO_MAPPING 1178 bool 1179 1180config MEMFD_CREATE 1181 bool "Enable memfd_create() system call" if EXPERT 1182 1183config SECRETMEM 1184 default y 1185 bool "Enable memfd_secret() system call" if EXPERT 1186 depends on ARCH_HAS_SET_DIRECT_MAP 1187 help 1188 Enable the memfd_secret() system call with the ability to create 1189 memory areas visible only in the context of the owning process and 1190 not mapped to other processes and other kernel page tables. 1191 1192config ANON_VMA_NAME 1193 bool "Anonymous VMA name support" 1194 depends on PROC_FS && ADVISE_SYSCALLS && MMU 1195 1196 help 1197 Allow naming anonymous virtual memory areas. 1198 1199 This feature allows assigning names to virtual memory areas. Assigned 1200 names can be later retrieved from /proc/pid/maps and /proc/pid/smaps 1201 and help identifying individual anonymous memory areas. 1202 Assigning a name to anonymous virtual memory area might prevent that 1203 area from being merged with adjacent virtual memory areas due to the 1204 difference in their name. 1205 1206config HAVE_ARCH_USERFAULTFD_WP 1207 bool 1208 help 1209 Arch has userfaultfd write protection support 1210 1211config HAVE_ARCH_USERFAULTFD_MINOR 1212 bool 1213 help 1214 Arch has userfaultfd minor fault support 1215 1216menuconfig USERFAULTFD 1217 bool "Enable userfaultfd() system call" 1218 depends on MMU 1219 help 1220 Enable the userfaultfd() system call that allows to intercept and 1221 handle page faults in userland. 1222 1223if USERFAULTFD 1224config PTE_MARKER_UFFD_WP 1225 bool "Userfaultfd write protection support for shmem/hugetlbfs" 1226 default y 1227 depends on HAVE_ARCH_USERFAULTFD_WP 1228 1229 help 1230 Allows to create marker PTEs for userfaultfd write protection 1231 purposes. It is required to enable userfaultfd write protection on 1232 file-backed memory types like shmem and hugetlbfs. 1233endif # USERFAULTFD 1234 1235# multi-gen LRU { 1236config LRU_GEN 1237 bool "Multi-Gen LRU" 1238 depends on MMU 1239 # make sure folio->flags has enough spare bits 1240 depends on 64BIT || !SPARSEMEM || SPARSEMEM_VMEMMAP 1241 help 1242 A high performance LRU implementation to overcommit memory. See 1243 Documentation/admin-guide/mm/multigen_lru.rst for details. 1244 1245config LRU_GEN_ENABLED 1246 bool "Enable by default" 1247 depends on LRU_GEN 1248 help 1249 This option enables the multi-gen LRU by default. 1250 1251config LRU_GEN_STATS 1252 bool "Full stats for debugging" 1253 depends on LRU_GEN 1254 help 1255 Do not enable this option unless you plan to look at historical stats 1256 from evicted generations for debugging purpose. 1257 1258 This option has a per-memcg and per-node memory overhead. 1259 1260config LRU_GEN_WALKS_MMU 1261 def_bool y 1262 depends on LRU_GEN && ARCH_HAS_HW_PTE_YOUNG 1263# } 1264 1265config ARCH_SUPPORTS_PER_VMA_LOCK 1266 def_bool n 1267 1268config PER_VMA_LOCK 1269 def_bool y 1270 depends on ARCH_SUPPORTS_PER_VMA_LOCK && MMU && SMP 1271 help 1272 Allow per-vma locking during page fault handling. 1273 1274 This feature allows locking each virtual memory area separately when 1275 handling page faults instead of taking mmap_lock. 1276 1277config LOCK_MM_AND_FIND_VMA 1278 bool 1279 depends on !STACK_GROWSUP 1280 1281config IOMMU_MM_DATA 1282 bool 1283 1284config EXECMEM 1285 bool 1286 1287config NUMA_MEMBLKS 1288 bool 1289 1290config NUMA_EMU 1291 bool "NUMA emulation" 1292 depends on NUMA_MEMBLKS 1293 help 1294 Enable NUMA emulation. A flat machine will be split 1295 into virtual nodes when booted with "numa=fake=N", where N is the 1296 number of nodes. This is only useful for debugging. 1297 1298source "mm/damon/Kconfig" 1299 1300endmenu 1301