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 150 bool "z3fold" 151 select Z3FOLD 152 help 153 Use the z3fold allocator as the default allocator. 154 155config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC 156 bool "zsmalloc" 157 select ZSMALLOC 158 help 159 Use the zsmalloc allocator as the default allocator. 160endchoice 161 162config ZSWAP_ZPOOL_DEFAULT 163 string 164 depends on ZSWAP 165 default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD 166 default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD 167 default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC 168 default "" 169 170config ZBUD 171 tristate "2:1 compression allocator (zbud)" 172 depends on ZSWAP 173 help 174 A special purpose allocator for storing compressed pages. 175 It is designed to store up to two compressed pages per physical 176 page. While this design limits storage density, it has simple and 177 deterministic reclaim properties that make it preferable to a higher 178 density approach when reclaim will be used. 179 180config Z3FOLD 181 tristate "3:1 compression allocator (z3fold)" 182 depends on ZSWAP 183 help 184 A special purpose allocator for storing compressed pages. 185 It is designed to store up to three compressed pages per physical 186 page. It is a ZBUD derivative so the simplicity and determinism are 187 still there. 188 189config ZSMALLOC 190 tristate 191 prompt "N:1 compression allocator (zsmalloc)" if ZSWAP 192 depends on MMU 193 help 194 zsmalloc is a slab-based memory allocator designed to store 195 pages of various compression levels efficiently. It achieves 196 the highest storage density with the least amount of fragmentation. 197 198config ZSMALLOC_STAT 199 bool "Export zsmalloc statistics" 200 depends on ZSMALLOC 201 select DEBUG_FS 202 help 203 This option enables code in the zsmalloc to collect various 204 statistics about what's happening in zsmalloc and exports that 205 information to userspace via debugfs. 206 If unsure, say N. 207 208config ZSMALLOC_CHAIN_SIZE 209 int "Maximum number of physical pages per-zspage" 210 default 8 211 range 4 16 212 depends on ZSMALLOC 213 help 214 This option sets the upper limit on the number of physical pages 215 that a zmalloc page (zspage) can consist of. The optimal zspage 216 chain size is calculated for each size class during the 217 initialization of the pool. 218 219 Changing this option can alter the characteristics of size classes, 220 such as the number of pages per zspage and the number of objects 221 per zspage. This can also result in different configurations of 222 the pool, as zsmalloc merges size classes with similar 223 characteristics. 224 225 For more information, see zsmalloc documentation. 226 227menu "Slab allocator options" 228 229config SLUB 230 def_bool y 231 232config SLUB_TINY 233 bool "Configure for minimal memory footprint" 234 depends on EXPERT 235 select SLAB_MERGE_DEFAULT 236 help 237 Configures the slab allocator in a way to achieve minimal memory 238 footprint, sacrificing scalability, debugging and other features. 239 This is intended only for the smallest system that had used the 240 SLOB allocator and is not recommended for systems with more than 241 16MB RAM. 242 243 If unsure, say N. 244 245config SLAB_MERGE_DEFAULT 246 bool "Allow slab caches to be merged" 247 default y 248 help 249 For reduced kernel memory fragmentation, slab caches can be 250 merged when they share the same size and other characteristics. 251 This carries a risk of kernel heap overflows being able to 252 overwrite objects from merged caches (and more easily control 253 cache layout), which makes such heap attacks easier to exploit 254 by attackers. By keeping caches unmerged, these kinds of exploits 255 can usually only damage objects in the same cache. To disable 256 merging at runtime, "slab_nomerge" can be passed on the kernel 257 command line. 258 259config SLAB_FREELIST_RANDOM 260 bool "Randomize slab freelist" 261 depends on !SLUB_TINY 262 help 263 Randomizes the freelist order used on creating new pages. This 264 security feature reduces the predictability of the kernel slab 265 allocator against heap overflows. 266 267config SLAB_FREELIST_HARDENED 268 bool "Harden slab freelist metadata" 269 depends on !SLUB_TINY 270 help 271 Many kernel heap attacks try to target slab cache metadata and 272 other infrastructure. This options makes minor performance 273 sacrifices to harden the kernel slab allocator against common 274 freelist exploit methods. 275 276config SLUB_STATS 277 default n 278 bool "Enable performance statistics" 279 depends on SYSFS && !SLUB_TINY 280 help 281 The statistics are useful to debug slab allocation behavior in 282 order find ways to optimize the allocator. This should never be 283 enabled for production use since keeping statistics slows down 284 the allocator by a few percentage points. The slabinfo command 285 supports the determination of the most active slabs to figure 286 out which slabs are relevant to a particular load. 287 Try running: slabinfo -DA 288 289config SLUB_CPU_PARTIAL 290 default y 291 depends on SMP && !SLUB_TINY 292 bool "Enable per cpu partial caches" 293 help 294 Per cpu partial caches accelerate objects allocation and freeing 295 that is local to a processor at the price of more indeterminism 296 in the latency of the free. On overflow these caches will be cleared 297 which requires the taking of locks that may cause latency spikes. 298 Typically one would choose no for a realtime system. 299 300config RANDOM_KMALLOC_CACHES 301 default n 302 depends on !SLUB_TINY 303 bool "Randomize slab caches for normal kmalloc" 304 help 305 A hardening feature that creates multiple copies of slab caches for 306 normal kmalloc allocation and makes kmalloc randomly pick one based 307 on code address, which makes the attackers more difficult to spray 308 vulnerable memory objects on the heap for the purpose of exploiting 309 memory vulnerabilities. 310 311 Currently the number of copies is set to 16, a reasonably large value 312 that effectively diverges the memory objects allocated for different 313 subsystems or modules into different caches, at the expense of a 314 limited degree of memory and CPU overhead that relates to hardware and 315 system workload. 316 317endmenu # Slab allocator options 318 319config SHUFFLE_PAGE_ALLOCATOR 320 bool "Page allocator randomization" 321 default SLAB_FREELIST_RANDOM && ACPI_NUMA 322 help 323 Randomization of the page allocator improves the average 324 utilization of a direct-mapped memory-side-cache. See section 325 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI 326 6.2a specification for an example of how a platform advertises 327 the presence of a memory-side-cache. There are also incidental 328 security benefits as it reduces the predictability of page 329 allocations to compliment SLAB_FREELIST_RANDOM, but the 330 default granularity of shuffling on the MAX_PAGE_ORDER i.e, 10th 331 order of pages is selected based on cache utilization benefits 332 on x86. 333 334 While the randomization improves cache utilization it may 335 negatively impact workloads on platforms without a cache. For 336 this reason, by default, the randomization is enabled only 337 after runtime detection of a direct-mapped memory-side-cache. 338 Otherwise, the randomization may be force enabled with the 339 'page_alloc.shuffle' kernel command line parameter. 340 341 Say Y if unsure. 342 343config COMPAT_BRK 344 bool "Disable heap randomization" 345 default y 346 help 347 Randomizing heap placement makes heap exploits harder, but it 348 also breaks ancient binaries (including anything libc5 based). 349 This option changes the bootup default to heap randomization 350 disabled, and can be overridden at runtime by setting 351 /proc/sys/kernel/randomize_va_space to 2. 352 353 On non-ancient distros (post-2000 ones) N is usually a safe choice. 354 355config MMAP_ALLOW_UNINITIALIZED 356 bool "Allow mmapped anonymous memory to be uninitialized" 357 depends on EXPERT && !MMU 358 default n 359 help 360 Normally, and according to the Linux spec, anonymous memory obtained 361 from mmap() has its contents cleared before it is passed to 362 userspace. Enabling this config option allows you to request that 363 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus 364 providing a huge performance boost. If this option is not enabled, 365 then the flag will be ignored. 366 367 This is taken advantage of by uClibc's malloc(), and also by 368 ELF-FDPIC binfmt's brk and stack allocator. 369 370 Because of the obvious security issues, this option should only be 371 enabled on embedded devices where you control what is run in 372 userspace. Since that isn't generally a problem on no-MMU systems, 373 it is normally safe to say Y here. 374 375 See Documentation/admin-guide/mm/nommu-mmap.rst for more information. 376 377config SELECT_MEMORY_MODEL 378 def_bool y 379 depends on ARCH_SELECT_MEMORY_MODEL 380 381choice 382 prompt "Memory model" 383 depends on SELECT_MEMORY_MODEL 384 default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT 385 default FLATMEM_MANUAL 386 help 387 This option allows you to change some of the ways that 388 Linux manages its memory internally. Most users will 389 only have one option here selected by the architecture 390 configuration. This is normal. 391 392config FLATMEM_MANUAL 393 bool "Flat Memory" 394 depends on !ARCH_SPARSEMEM_ENABLE || ARCH_FLATMEM_ENABLE 395 help 396 This option is best suited for non-NUMA systems with 397 flat address space. The FLATMEM is the most efficient 398 system in terms of performance and resource consumption 399 and it is the best option for smaller systems. 400 401 For systems that have holes in their physical address 402 spaces and for features like NUMA and memory hotplug, 403 choose "Sparse Memory". 404 405 If unsure, choose this option (Flat Memory) over any other. 406 407config SPARSEMEM_MANUAL 408 bool "Sparse Memory" 409 depends on ARCH_SPARSEMEM_ENABLE 410 help 411 This will be the only option for some systems, including 412 memory hot-plug systems. This is normal. 413 414 This option provides efficient support for systems with 415 holes is their physical address space and allows memory 416 hot-plug and hot-remove. 417 418 If unsure, choose "Flat Memory" over this option. 419 420endchoice 421 422config SPARSEMEM 423 def_bool y 424 depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL 425 426config FLATMEM 427 def_bool y 428 depends on !SPARSEMEM || FLATMEM_MANUAL 429 430# 431# SPARSEMEM_EXTREME (which is the default) does some bootmem 432# allocations when sparse_init() is called. If this cannot 433# be done on your architecture, select this option. However, 434# statically allocating the mem_section[] array can potentially 435# consume vast quantities of .bss, so be careful. 436# 437# This option will also potentially produce smaller runtime code 438# with gcc 3.4 and later. 439# 440config SPARSEMEM_STATIC 441 bool 442 443# 444# Architecture platforms which require a two level mem_section in SPARSEMEM 445# must select this option. This is usually for architecture platforms with 446# an extremely sparse physical address space. 447# 448config SPARSEMEM_EXTREME 449 def_bool y 450 depends on SPARSEMEM && !SPARSEMEM_STATIC 451 452config SPARSEMEM_VMEMMAP_ENABLE 453 bool 454 455config SPARSEMEM_VMEMMAP 456 bool "Sparse Memory virtual memmap" 457 depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE 458 default y 459 help 460 SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise 461 pfn_to_page and page_to_pfn operations. This is the most 462 efficient option when sufficient kernel resources are available. 463# 464# Select this config option from the architecture Kconfig, if it is preferred 465# to enable the feature of HugeTLB/dev_dax vmemmap optimization. 466# 467config ARCH_WANT_OPTIMIZE_DAX_VMEMMAP 468 bool 469 470config ARCH_WANT_OPTIMIZE_HUGETLB_VMEMMAP 471 bool 472 473config HAVE_MEMBLOCK_PHYS_MAP 474 bool 475 476config HAVE_FAST_GUP 477 depends on MMU 478 bool 479 480# Don't discard allocated memory used to track "memory" and "reserved" memblocks 481# after early boot, so it can still be used to test for validity of memory. 482# Also, memblocks are updated with memory hot(un)plug. 483config ARCH_KEEP_MEMBLOCK 484 bool 485 486# Keep arch NUMA mapping infrastructure post-init. 487config NUMA_KEEP_MEMINFO 488 bool 489 490config MEMORY_ISOLATION 491 bool 492 493# IORESOURCE_SYSTEM_RAM regions in the kernel resource tree that are marked 494# IORESOURCE_EXCLUSIVE cannot be mapped to user space, for example, via 495# /dev/mem. 496config EXCLUSIVE_SYSTEM_RAM 497 def_bool y 498 depends on !DEVMEM || STRICT_DEVMEM 499 500# 501# Only be set on architectures that have completely implemented memory hotplug 502# feature. If you are not sure, don't touch it. 503# 504config HAVE_BOOTMEM_INFO_NODE 505 def_bool n 506 507config ARCH_ENABLE_MEMORY_HOTPLUG 508 bool 509 510config ARCH_ENABLE_MEMORY_HOTREMOVE 511 bool 512 513# eventually, we can have this option just 'select SPARSEMEM' 514menuconfig MEMORY_HOTPLUG 515 bool "Memory hotplug" 516 select MEMORY_ISOLATION 517 depends on SPARSEMEM 518 depends on ARCH_ENABLE_MEMORY_HOTPLUG 519 depends on 64BIT 520 select NUMA_KEEP_MEMINFO if NUMA 521 522if MEMORY_HOTPLUG 523 524config MEMORY_HOTPLUG_DEFAULT_ONLINE 525 bool "Online the newly added memory blocks by default" 526 depends on MEMORY_HOTPLUG 527 help 528 This option sets the default policy setting for memory hotplug 529 onlining policy (/sys/devices/system/memory/auto_online_blocks) which 530 determines what happens to newly added memory regions. Policy setting 531 can always be changed at runtime. 532 See Documentation/admin-guide/mm/memory-hotplug.rst for more information. 533 534 Say Y here if you want all hot-plugged memory blocks to appear in 535 'online' state by default. 536 Say N here if you want the default policy to keep all hot-plugged 537 memory blocks in 'offline' state. 538 539config MEMORY_HOTREMOVE 540 bool "Allow for memory hot remove" 541 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64) 542 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE 543 depends on MIGRATION 544 545config MHP_MEMMAP_ON_MEMORY 546 def_bool y 547 depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP 548 depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE 549 550endif # MEMORY_HOTPLUG 551 552config ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE 553 bool 554 555# Heavily threaded applications may benefit from splitting the mm-wide 556# page_table_lock, so that faults on different parts of the user address 557# space can be handled with less contention: split it at this NR_CPUS. 558# Default to 4 for wider testing, though 8 might be more appropriate. 559# ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock. 560# PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes. 561# SPARC32 allocates multiple pte tables within a single page, and therefore 562# a per-page lock leads to problems when multiple tables need to be locked 563# at the same time (e.g. copy_page_range()). 564# DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page. 565# 566config SPLIT_PTLOCK_CPUS 567 int 568 default "999999" if !MMU 569 default "999999" if ARM && !CPU_CACHE_VIPT 570 default "999999" if PARISC && !PA20 571 default "999999" if SPARC32 572 default "4" 573 574config ARCH_ENABLE_SPLIT_PMD_PTLOCK 575 bool 576 577# 578# support for memory balloon 579config MEMORY_BALLOON 580 bool 581 582# 583# support for memory balloon compaction 584config BALLOON_COMPACTION 585 bool "Allow for balloon memory compaction/migration" 586 def_bool y 587 depends on COMPACTION && MEMORY_BALLOON 588 help 589 Memory fragmentation introduced by ballooning might reduce 590 significantly the number of 2MB contiguous memory blocks that can be 591 used within a guest, thus imposing performance penalties associated 592 with the reduced number of transparent huge pages that could be used 593 by the guest workload. Allowing the compaction & migration for memory 594 pages enlisted as being part of memory balloon devices avoids the 595 scenario aforementioned and helps improving memory defragmentation. 596 597# 598# support for memory compaction 599config COMPACTION 600 bool "Allow for memory compaction" 601 def_bool y 602 select MIGRATION 603 depends on MMU 604 help 605 Compaction is the only memory management component to form 606 high order (larger physically contiguous) memory blocks 607 reliably. The page allocator relies on compaction heavily and 608 the lack of the feature can lead to unexpected OOM killer 609 invocations for high order memory requests. You shouldn't 610 disable this option unless there really is a strong reason for 611 it and then we would be really interested to hear about that at 612 linux-mm@kvack.org. 613 614config COMPACT_UNEVICTABLE_DEFAULT 615 int 616 depends on COMPACTION 617 default 0 if PREEMPT_RT 618 default 1 619 620# 621# support for free page reporting 622config PAGE_REPORTING 623 bool "Free page reporting" 624 def_bool n 625 help 626 Free page reporting allows for the incremental acquisition of 627 free pages from the buddy allocator for the purpose of reporting 628 those pages to another entity, such as a hypervisor, so that the 629 memory can be freed within the host for other uses. 630 631# 632# support for page migration 633# 634config MIGRATION 635 bool "Page migration" 636 def_bool y 637 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU 638 help 639 Allows the migration of the physical location of pages of processes 640 while the virtual addresses are not changed. This is useful in 641 two situations. The first is on NUMA systems to put pages nearer 642 to the processors accessing. The second is when allocating huge 643 pages as migration can relocate pages to satisfy a huge page 644 allocation instead of reclaiming. 645 646config DEVICE_MIGRATION 647 def_bool MIGRATION && ZONE_DEVICE 648 649config ARCH_ENABLE_HUGEPAGE_MIGRATION 650 bool 651 652config ARCH_ENABLE_THP_MIGRATION 653 bool 654 655config HUGETLB_PAGE_SIZE_VARIABLE 656 def_bool n 657 help 658 Allows the pageblock_order value to be dynamic instead of just standard 659 HUGETLB_PAGE_ORDER when there are multiple HugeTLB page sizes available 660 on a platform. 661 662 Note that the pageblock_order cannot exceed MAX_PAGE_ORDER and will be 663 clamped down to MAX_PAGE_ORDER. 664 665config CONTIG_ALLOC 666 def_bool (MEMORY_ISOLATION && COMPACTION) || CMA 667 668config PCP_BATCH_SCALE_MAX 669 int "Maximum scale factor of PCP (Per-CPU pageset) batch allocate/free" 670 default 5 671 range 0 6 672 help 673 In page allocator, PCP (Per-CPU pageset) is refilled and drained in 674 batches. The batch number is scaled automatically to improve page 675 allocation/free throughput. But too large scale factor may hurt 676 latency. This option sets the upper limit of scale factor to limit 677 the maximum latency. 678 679config PHYS_ADDR_T_64BIT 680 def_bool 64BIT 681 682config BOUNCE 683 bool "Enable bounce buffers" 684 default y 685 depends on BLOCK && MMU && HIGHMEM 686 help 687 Enable bounce buffers for devices that cannot access the full range of 688 memory available to the CPU. Enabled by default when HIGHMEM is 689 selected, but you may say n to override this. 690 691config MMU_NOTIFIER 692 bool 693 select INTERVAL_TREE 694 695config KSM 696 bool "Enable KSM for page merging" 697 depends on MMU 698 select XXHASH 699 help 700 Enable Kernel Samepage Merging: KSM periodically scans those areas 701 of an application's address space that an app has advised may be 702 mergeable. When it finds pages of identical content, it replaces 703 the many instances by a single page with that content, so 704 saving memory until one or another app needs to modify the content. 705 Recommended for use with KVM, or with other duplicative applications. 706 See Documentation/mm/ksm.rst for more information: KSM is inactive 707 until a program has madvised that an area is MADV_MERGEABLE, and 708 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). 709 710config DEFAULT_MMAP_MIN_ADDR 711 int "Low address space to protect from user allocation" 712 depends on MMU 713 default 4096 714 help 715 This is the portion of low virtual memory which should be protected 716 from userspace allocation. Keeping a user from writing to low pages 717 can help reduce the impact of kernel NULL pointer bugs. 718 719 For most ppc64 and x86 users with lots of address space 720 a value of 65536 is reasonable and should cause no problems. 721 On arm and other archs it should not be higher than 32768. 722 Programs which use vm86 functionality or have some need to map 723 this low address space will need CAP_SYS_RAWIO or disable this 724 protection by setting the value to 0. 725 726 This value can be changed after boot using the 727 /proc/sys/vm/mmap_min_addr tunable. 728 729config ARCH_SUPPORTS_MEMORY_FAILURE 730 bool 731 732config MEMORY_FAILURE 733 depends on MMU 734 depends on ARCH_SUPPORTS_MEMORY_FAILURE 735 bool "Enable recovery from hardware memory errors" 736 select MEMORY_ISOLATION 737 select RAS 738 help 739 Enables code to recover from some memory failures on systems 740 with MCA recovery. This allows a system to continue running 741 even when some of its memory has uncorrected errors. This requires 742 special hardware support and typically ECC memory. 743 744config HWPOISON_INJECT 745 tristate "HWPoison pages injector" 746 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS 747 select PROC_PAGE_MONITOR 748 749config NOMMU_INITIAL_TRIM_EXCESS 750 int "Turn on mmap() excess space trimming before booting" 751 depends on !MMU 752 default 1 753 help 754 The NOMMU mmap() frequently needs to allocate large contiguous chunks 755 of memory on which to store mappings, but it can only ask the system 756 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently 757 more than it requires. To deal with this, mmap() is able to trim off 758 the excess and return it to the allocator. 759 760 If trimming is enabled, the excess is trimmed off and returned to the 761 system allocator, which can cause extra fragmentation, particularly 762 if there are a lot of transient processes. 763 764 If trimming is disabled, the excess is kept, but not used, which for 765 long-term mappings means that the space is wasted. 766 767 Trimming can be dynamically controlled through a sysctl option 768 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of 769 excess pages there must be before trimming should occur, or zero if 770 no trimming is to occur. 771 772 This option specifies the initial value of this option. The default 773 of 1 says that all excess pages should be trimmed. 774 775 See Documentation/admin-guide/mm/nommu-mmap.rst for more information. 776 777config ARCH_WANT_GENERAL_HUGETLB 778 bool 779 780config ARCH_WANTS_THP_SWAP 781 def_bool n 782 783menuconfig TRANSPARENT_HUGEPAGE 784 bool "Transparent Hugepage Support" 785 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE && !PREEMPT_RT 786 select COMPACTION 787 select XARRAY_MULTI 788 help 789 Transparent Hugepages allows the kernel to use huge pages and 790 huge tlb transparently to the applications whenever possible. 791 This feature can improve computing performance to certain 792 applications by speeding up page faults during memory 793 allocation, by reducing the number of tlb misses and by speeding 794 up the pagetable walking. 795 796 If memory constrained on embedded, you may want to say N. 797 798if TRANSPARENT_HUGEPAGE 799 800choice 801 prompt "Transparent Hugepage Support sysfs defaults" 802 depends on TRANSPARENT_HUGEPAGE 803 default TRANSPARENT_HUGEPAGE_ALWAYS 804 help 805 Selects the sysfs defaults for Transparent Hugepage Support. 806 807 config TRANSPARENT_HUGEPAGE_ALWAYS 808 bool "always" 809 help 810 Enabling Transparent Hugepage always, can increase the 811 memory footprint of applications without a guaranteed 812 benefit but it will work automatically for all applications. 813 814 config TRANSPARENT_HUGEPAGE_MADVISE 815 bool "madvise" 816 help 817 Enabling Transparent Hugepage madvise, will only provide a 818 performance improvement benefit to the applications using 819 madvise(MADV_HUGEPAGE) but it won't risk to increase the 820 memory footprint of applications without a guaranteed 821 benefit. 822 823 config TRANSPARENT_HUGEPAGE_NEVER 824 bool "never" 825 help 826 Disable Transparent Hugepage by default. It can still be 827 enabled at runtime via sysfs. 828endchoice 829 830config THP_SWAP 831 def_bool y 832 depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP && 64BIT 833 help 834 Swap transparent huge pages in one piece, without splitting. 835 XXX: For now, swap cluster backing transparent huge page 836 will be split after swapout. 837 838 For selection by architectures with reasonable THP sizes. 839 840config READ_ONLY_THP_FOR_FS 841 bool "Read-only THP for filesystems (EXPERIMENTAL)" 842 depends on TRANSPARENT_HUGEPAGE && SHMEM 843 844 help 845 Allow khugepaged to put read-only file-backed pages in THP. 846 847 This is marked experimental because it is a new feature. Write 848 support of file THPs will be developed in the next few release 849 cycles. 850 851endif # TRANSPARENT_HUGEPAGE 852 853# 854# UP and nommu archs use km based percpu allocator 855# 856config NEED_PER_CPU_KM 857 depends on !SMP || !MMU 858 bool 859 default y 860 861config NEED_PER_CPU_EMBED_FIRST_CHUNK 862 bool 863 864config NEED_PER_CPU_PAGE_FIRST_CHUNK 865 bool 866 867config USE_PERCPU_NUMA_NODE_ID 868 bool 869 870config HAVE_SETUP_PER_CPU_AREA 871 bool 872 873config CMA 874 bool "Contiguous Memory Allocator" 875 depends on MMU 876 select MIGRATION 877 select MEMORY_ISOLATION 878 help 879 This enables the Contiguous Memory Allocator which allows other 880 subsystems to allocate big physically-contiguous blocks of memory. 881 CMA reserves a region of memory and allows only movable pages to 882 be allocated from it. This way, the kernel can use the memory for 883 pagecache and when a subsystem requests for contiguous area, the 884 allocated pages are migrated away to serve the contiguous request. 885 886 If unsure, say "n". 887 888config CMA_DEBUGFS 889 bool "CMA debugfs interface" 890 depends on CMA && DEBUG_FS 891 help 892 Turns on the DebugFS interface for CMA. 893 894config CMA_SYSFS 895 bool "CMA information through sysfs interface" 896 depends on CMA && SYSFS 897 help 898 This option exposes some sysfs attributes to get information 899 from CMA. 900 901config CMA_AREAS 902 int "Maximum count of the CMA areas" 903 depends on CMA 904 default 20 if NUMA 905 default 8 906 help 907 CMA allows to create CMA areas for particular purpose, mainly, 908 used as device private area. This parameter sets the maximum 909 number of CMA area in the system. 910 911 If unsure, leave the default value "8" in UMA and "20" in NUMA. 912 913config MEM_SOFT_DIRTY 914 bool "Track memory changes" 915 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS 916 select PROC_PAGE_MONITOR 917 help 918 This option enables memory changes tracking by introducing a 919 soft-dirty bit on pte-s. This bit it set when someone writes 920 into a page just as regular dirty bit, but unlike the latter 921 it can be cleared by hands. 922 923 See Documentation/admin-guide/mm/soft-dirty.rst for more details. 924 925config GENERIC_EARLY_IOREMAP 926 bool 927 928config STACK_MAX_DEFAULT_SIZE_MB 929 int "Default maximum user stack size for 32-bit processes (MB)" 930 default 100 931 range 8 2048 932 depends on STACK_GROWSUP && (!64BIT || COMPAT) 933 help 934 This is the maximum stack size in Megabytes in the VM layout of 32-bit 935 user processes when the stack grows upwards (currently only on parisc 936 arch) when the RLIMIT_STACK hard limit is unlimited. 937 938 A sane initial value is 100 MB. 939 940config DEFERRED_STRUCT_PAGE_INIT 941 bool "Defer initialisation of struct pages to kthreads" 942 depends on SPARSEMEM 943 depends on !NEED_PER_CPU_KM 944 depends on 64BIT 945 select PADATA 946 help 947 Ordinarily all struct pages are initialised during early boot in a 948 single thread. On very large machines this can take a considerable 949 amount of time. If this option is set, large machines will bring up 950 a subset of memmap at boot and then initialise the rest in parallel. 951 This has a potential performance impact on tasks running early in the 952 lifetime of the system until these kthreads finish the 953 initialisation. 954 955config PAGE_IDLE_FLAG 956 bool 957 select PAGE_EXTENSION if !64BIT 958 help 959 This adds PG_idle and PG_young flags to 'struct page'. PTE Accessed 960 bit writers can set the state of the bit in the flags so that PTE 961 Accessed bit readers may avoid disturbance. 962 963config IDLE_PAGE_TRACKING 964 bool "Enable idle page tracking" 965 depends on SYSFS && MMU 966 select PAGE_IDLE_FLAG 967 help 968 This feature allows to estimate the amount of user pages that have 969 not been touched during a given period of time. This information can 970 be useful to tune memory cgroup limits and/or for job placement 971 within a compute cluster. 972 973 See Documentation/admin-guide/mm/idle_page_tracking.rst for 974 more details. 975 976# Architectures which implement cpu_dcache_is_aliasing() to query 977# whether the data caches are aliased (VIVT or VIPT with dcache 978# aliasing) need to select this. 979config ARCH_HAS_CPU_CACHE_ALIASING 980 bool 981 982config ARCH_HAS_CACHE_LINE_SIZE 983 bool 984 985config ARCH_HAS_CURRENT_STACK_POINTER 986 bool 987 help 988 In support of HARDENED_USERCOPY performing stack variable lifetime 989 checking, an architecture-agnostic way to find the stack pointer 990 is needed. Once an architecture defines an unsigned long global 991 register alias named "current_stack_pointer", this config can be 992 selected. 993 994config ARCH_HAS_PTE_DEVMAP 995 bool 996 997config ARCH_HAS_ZONE_DMA_SET 998 bool 999 1000config ZONE_DMA 1001 bool "Support DMA zone" if ARCH_HAS_ZONE_DMA_SET 1002 default y if ARM64 || X86 1003 1004config ZONE_DMA32 1005 bool "Support DMA32 zone" if ARCH_HAS_ZONE_DMA_SET 1006 depends on !X86_32 1007 default y if ARM64 1008 1009config ZONE_DEVICE 1010 bool "Device memory (pmem, HMM, etc...) hotplug support" 1011 depends on MEMORY_HOTPLUG 1012 depends on MEMORY_HOTREMOVE 1013 depends on SPARSEMEM_VMEMMAP 1014 depends on ARCH_HAS_PTE_DEVMAP 1015 select XARRAY_MULTI 1016 1017 help 1018 Device memory hotplug support allows for establishing pmem, 1019 or other device driver discovered memory regions, in the 1020 memmap. This allows pfn_to_page() lookups of otherwise 1021 "device-physical" addresses which is needed for using a DAX 1022 mapping in an O_DIRECT operation, among other things. 1023 1024 If FS_DAX is enabled, then say Y. 1025 1026# 1027# Helpers to mirror range of the CPU page tables of a process into device page 1028# tables. 1029# 1030config HMM_MIRROR 1031 bool 1032 depends on MMU 1033 1034config GET_FREE_REGION 1035 depends on SPARSEMEM 1036 bool 1037 1038config DEVICE_PRIVATE 1039 bool "Unaddressable device memory (GPU memory, ...)" 1040 depends on ZONE_DEVICE 1041 select GET_FREE_REGION 1042 1043 help 1044 Allows creation of struct pages to represent unaddressable device 1045 memory; i.e., memory that is only accessible from the device (or 1046 group of devices). You likely also want to select HMM_MIRROR. 1047 1048config VMAP_PFN 1049 bool 1050 1051config ARCH_USES_HIGH_VMA_FLAGS 1052 bool 1053config ARCH_HAS_PKEYS 1054 bool 1055 1056config ARCH_USES_PG_ARCH_X 1057 bool 1058 help 1059 Enable the definition of PG_arch_x page flags with x > 1. Only 1060 suitable for 64-bit architectures with CONFIG_FLATMEM or 1061 CONFIG_SPARSEMEM_VMEMMAP enabled, otherwise there may not be 1062 enough room for additional bits in page->flags. 1063 1064config VM_EVENT_COUNTERS 1065 default y 1066 bool "Enable VM event counters for /proc/vmstat" if EXPERT 1067 help 1068 VM event counters are needed for event counts to be shown. 1069 This option allows the disabling of the VM event counters 1070 on EXPERT systems. /proc/vmstat will only show page counts 1071 if VM event counters are disabled. 1072 1073config PERCPU_STATS 1074 bool "Collect percpu memory statistics" 1075 help 1076 This feature collects and exposes statistics via debugfs. The 1077 information includes global and per chunk statistics, which can 1078 be used to help understand percpu memory usage. 1079 1080config GUP_TEST 1081 bool "Enable infrastructure for get_user_pages()-related unit tests" 1082 depends on DEBUG_FS 1083 help 1084 Provides /sys/kernel/debug/gup_test, which in turn provides a way 1085 to make ioctl calls that can launch kernel-based unit tests for 1086 the get_user_pages*() and pin_user_pages*() family of API calls. 1087 1088 These tests include benchmark testing of the _fast variants of 1089 get_user_pages*() and pin_user_pages*(), as well as smoke tests of 1090 the non-_fast variants. 1091 1092 There is also a sub-test that allows running dump_page() on any 1093 of up to eight pages (selected by command line args) within the 1094 range of user-space addresses. These pages are either pinned via 1095 pin_user_pages*(), or pinned via get_user_pages*(), as specified 1096 by other command line arguments. 1097 1098 See tools/testing/selftests/mm/gup_test.c 1099 1100comment "GUP_TEST needs to have DEBUG_FS enabled" 1101 depends on !GUP_TEST && !DEBUG_FS 1102 1103config GUP_GET_PXX_LOW_HIGH 1104 bool 1105 1106config DMAPOOL_TEST 1107 tristate "Enable a module to run time tests on dma_pool" 1108 depends on HAS_DMA 1109 help 1110 Provides a test module that will allocate and free many blocks of 1111 various sizes and report how long it takes. This is intended to 1112 provide a consistent way to measure how changes to the 1113 dma_pool_alloc/free routines affect performance. 1114 1115config ARCH_HAS_PTE_SPECIAL 1116 bool 1117 1118# 1119# Some architectures require a special hugepage directory format that is 1120# required to support multiple hugepage sizes. For example a4fe3ce76 1121# "powerpc/mm: Allow more flexible layouts for hugepage pagetables" 1122# introduced it on powerpc. This allows for a more flexible hugepage 1123# pagetable layouts. 1124# 1125config ARCH_HAS_HUGEPD 1126 bool 1127 1128config MAPPING_DIRTY_HELPERS 1129 bool 1130 1131config KMAP_LOCAL 1132 bool 1133 1134config KMAP_LOCAL_NON_LINEAR_PTE_ARRAY 1135 bool 1136 1137# struct io_mapping based helper. Selected by drivers that need them 1138config IO_MAPPING 1139 bool 1140 1141config MEMFD_CREATE 1142 bool "Enable memfd_create() system call" if EXPERT 1143 1144config SECRETMEM 1145 default y 1146 bool "Enable memfd_secret() system call" if EXPERT 1147 depends on ARCH_HAS_SET_DIRECT_MAP 1148 help 1149 Enable the memfd_secret() system call with the ability to create 1150 memory areas visible only in the context of the owning process and 1151 not mapped to other processes and other kernel page tables. 1152 1153config ANON_VMA_NAME 1154 bool "Anonymous VMA name support" 1155 depends on PROC_FS && ADVISE_SYSCALLS && MMU 1156 1157 help 1158 Allow naming anonymous virtual memory areas. 1159 1160 This feature allows assigning names to virtual memory areas. Assigned 1161 names can be later retrieved from /proc/pid/maps and /proc/pid/smaps 1162 and help identifying individual anonymous memory areas. 1163 Assigning a name to anonymous virtual memory area might prevent that 1164 area from being merged with adjacent virtual memory areas due to the 1165 difference in their name. 1166 1167config HAVE_ARCH_USERFAULTFD_WP 1168 bool 1169 help 1170 Arch has userfaultfd write protection support 1171 1172config HAVE_ARCH_USERFAULTFD_MINOR 1173 bool 1174 help 1175 Arch has userfaultfd minor fault support 1176 1177menuconfig USERFAULTFD 1178 bool "Enable userfaultfd() system call" 1179 depends on MMU 1180 help 1181 Enable the userfaultfd() system call that allows to intercept and 1182 handle page faults in userland. 1183 1184if USERFAULTFD 1185config PTE_MARKER_UFFD_WP 1186 bool "Userfaultfd write protection support for shmem/hugetlbfs" 1187 default y 1188 depends on HAVE_ARCH_USERFAULTFD_WP 1189 1190 help 1191 Allows to create marker PTEs for userfaultfd write protection 1192 purposes. It is required to enable userfaultfd write protection on 1193 file-backed memory types like shmem and hugetlbfs. 1194endif # USERFAULTFD 1195 1196# multi-gen LRU { 1197config LRU_GEN 1198 bool "Multi-Gen LRU" 1199 depends on MMU 1200 # make sure folio->flags has enough spare bits 1201 depends on 64BIT || !SPARSEMEM || SPARSEMEM_VMEMMAP 1202 help 1203 A high performance LRU implementation to overcommit memory. See 1204 Documentation/admin-guide/mm/multigen_lru.rst for details. 1205 1206config LRU_GEN_ENABLED 1207 bool "Enable by default" 1208 depends on LRU_GEN 1209 help 1210 This option enables the multi-gen LRU by default. 1211 1212config LRU_GEN_STATS 1213 bool "Full stats for debugging" 1214 depends on LRU_GEN 1215 help 1216 Do not enable this option unless you plan to look at historical stats 1217 from evicted generations for debugging purpose. 1218 1219 This option has a per-memcg and per-node memory overhead. 1220 1221config LRU_GEN_WALKS_MMU 1222 def_bool y 1223 depends on LRU_GEN && ARCH_HAS_HW_PTE_YOUNG 1224# } 1225 1226config ARCH_SUPPORTS_PER_VMA_LOCK 1227 def_bool n 1228 1229config PER_VMA_LOCK 1230 def_bool y 1231 depends on ARCH_SUPPORTS_PER_VMA_LOCK && MMU && SMP 1232 help 1233 Allow per-vma locking during page fault handling. 1234 1235 This feature allows locking each virtual memory area separately when 1236 handling page faults instead of taking mmap_lock. 1237 1238config LOCK_MM_AND_FIND_VMA 1239 bool 1240 depends on !STACK_GROWSUP 1241 1242config IOMMU_MM_DATA 1243 bool 1244 1245source "mm/damon/Kconfig" 1246 1247endmenu 1248