Thomas Gleixner | ec8f24b | 2019-05-19 13:07:45 +0100 | [diff] [blame] | 1 | # SPDX-License-Identifier: GPL-2.0-only |
Christoph Hellwig | 59e0b52 | 2018-07-31 13:39:35 +0200 | [diff] [blame] | 2 | |
| 3 | menu "Memory Management options" |
| 4 | |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 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 | # |
| 9 | config ARCH_NO_SWAP |
| 10 | bool |
| 11 | |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 12 | config ZPOOL |
| 13 | bool |
| 14 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 15 | menuconfig SWAP |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 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 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 25 | config ZSWAP |
David Heidelberg | fcab9b4 | 2022-08-23 17:20:33 +0200 | [diff] [blame] | 26 | bool "Compressed cache for swap pages" |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 27 | depends on SWAP |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 28 | select CRYPTO |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 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, |
Sophia Gabriella | 1a44131 | 2022-07-28 16:51:39 +0000 | [diff] [blame] | 35 | in the case where decompressing from RAM is faster than swap device |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 36 | reads, can also improve workload performance. |
| 37 | |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 38 | config 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 | |
Yosry Ahmed | b9c91c4 | 2023-06-07 19:51:43 +0000 | [diff] [blame] | 48 | config ZSWAP_EXCLUSIVE_LOADS_DEFAULT_ON |
| 49 | bool "Invalidate zswap entries when pages are loaded" |
| 50 | depends on ZSWAP |
| 51 | help |
| 52 | If selected, exclusive loads for zswap will be enabled at boot, |
| 53 | otherwise it will be disabled. |
| 54 | |
| 55 | If exclusive loads are enabled, when a page is loaded from zswap, |
| 56 | the zswap entry is invalidated at once, as opposed to leaving it |
| 57 | in zswap until the swap entry is freed. |
| 58 | |
| 59 | This avoids having two copies of the same page in memory |
| 60 | (compressed and uncompressed) after faulting in a page from zswap. |
| 61 | The cost is that if the page was never dirtied and needs to be |
| 62 | swapped out again, it will be re-compressed. |
| 63 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 64 | choice |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 65 | prompt "Default compressor" |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 66 | depends on ZSWAP |
| 67 | default ZSWAP_COMPRESSOR_DEFAULT_LZO |
| 68 | help |
| 69 | Selects the default compression algorithm for the compressed cache |
| 70 | for swap pages. |
| 71 | |
| 72 | For an overview what kind of performance can be expected from |
| 73 | a particular compression algorithm please refer to the benchmarks |
| 74 | available at the following LWN page: |
| 75 | https://lwn.net/Articles/751795/ |
| 76 | |
| 77 | If in doubt, select 'LZO'. |
| 78 | |
| 79 | The selection made here can be overridden by using the kernel |
| 80 | command line 'zswap.compressor=' option. |
| 81 | |
| 82 | config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE |
| 83 | bool "Deflate" |
| 84 | select CRYPTO_DEFLATE |
| 85 | help |
| 86 | Use the Deflate algorithm as the default compression algorithm. |
| 87 | |
| 88 | config ZSWAP_COMPRESSOR_DEFAULT_LZO |
| 89 | bool "LZO" |
| 90 | select CRYPTO_LZO |
| 91 | help |
| 92 | Use the LZO algorithm as the default compression algorithm. |
| 93 | |
| 94 | config ZSWAP_COMPRESSOR_DEFAULT_842 |
| 95 | bool "842" |
| 96 | select CRYPTO_842 |
| 97 | help |
| 98 | Use the 842 algorithm as the default compression algorithm. |
| 99 | |
| 100 | config ZSWAP_COMPRESSOR_DEFAULT_LZ4 |
| 101 | bool "LZ4" |
| 102 | select CRYPTO_LZ4 |
| 103 | help |
| 104 | Use the LZ4 algorithm as the default compression algorithm. |
| 105 | |
| 106 | config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC |
| 107 | bool "LZ4HC" |
| 108 | select CRYPTO_LZ4HC |
| 109 | help |
| 110 | Use the LZ4HC algorithm as the default compression algorithm. |
| 111 | |
| 112 | config ZSWAP_COMPRESSOR_DEFAULT_ZSTD |
| 113 | bool "zstd" |
| 114 | select CRYPTO_ZSTD |
| 115 | help |
| 116 | Use the zstd algorithm as the default compression algorithm. |
| 117 | endchoice |
| 118 | |
| 119 | config ZSWAP_COMPRESSOR_DEFAULT |
| 120 | string |
| 121 | depends on ZSWAP |
| 122 | default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE |
| 123 | default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO |
| 124 | default "842" if ZSWAP_COMPRESSOR_DEFAULT_842 |
| 125 | default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4 |
| 126 | default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC |
| 127 | default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD |
| 128 | default "" |
| 129 | |
| 130 | choice |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 131 | prompt "Default allocator" |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 132 | depends on ZSWAP |
| 133 | default ZSWAP_ZPOOL_DEFAULT_ZBUD |
| 134 | help |
| 135 | Selects the default allocator for the compressed cache for |
| 136 | swap pages. |
| 137 | The default is 'zbud' for compatibility, however please do |
| 138 | read the description of each of the allocators below before |
| 139 | making a right choice. |
| 140 | |
| 141 | The selection made here can be overridden by using the kernel |
| 142 | command line 'zswap.zpool=' option. |
| 143 | |
| 144 | config ZSWAP_ZPOOL_DEFAULT_ZBUD |
| 145 | bool "zbud" |
| 146 | select ZBUD |
| 147 | help |
| 148 | Use the zbud allocator as the default allocator. |
| 149 | |
| 150 | config ZSWAP_ZPOOL_DEFAULT_Z3FOLD |
| 151 | bool "z3fold" |
| 152 | select Z3FOLD |
| 153 | help |
| 154 | Use the z3fold allocator as the default allocator. |
| 155 | |
| 156 | config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC |
| 157 | bool "zsmalloc" |
| 158 | select ZSMALLOC |
| 159 | help |
| 160 | Use the zsmalloc allocator as the default allocator. |
| 161 | endchoice |
| 162 | |
| 163 | config ZSWAP_ZPOOL_DEFAULT |
| 164 | string |
| 165 | depends on ZSWAP |
| 166 | default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD |
| 167 | default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD |
| 168 | default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC |
| 169 | default "" |
| 170 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 171 | config ZBUD |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 172 | tristate "2:1 compression allocator (zbud)" |
| 173 | depends on ZSWAP |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 174 | help |
| 175 | A special purpose allocator for storing compressed pages. |
| 176 | It is designed to store up to two compressed pages per physical |
| 177 | page. While this design limits storage density, it has simple and |
| 178 | deterministic reclaim properties that make it preferable to a higher |
| 179 | density approach when reclaim will be used. |
| 180 | |
| 181 | config Z3FOLD |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 182 | tristate "3:1 compression allocator (z3fold)" |
| 183 | depends on ZSWAP |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 184 | help |
| 185 | A special purpose allocator for storing compressed pages. |
| 186 | It is designed to store up to three compressed pages per physical |
| 187 | page. It is a ZBUD derivative so the simplicity and determinism are |
| 188 | still there. |
| 189 | |
| 190 | config ZSMALLOC |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 191 | tristate |
| 192 | prompt "N:1 compression allocator (zsmalloc)" if ZSWAP |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 193 | depends on MMU |
| 194 | help |
| 195 | zsmalloc is a slab-based memory allocator designed to store |
Johannes Weiner | b3fbd58 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 196 | pages of various compression levels efficiently. It achieves |
| 197 | the highest storage density with the least amount of fragmentation. |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 198 | |
| 199 | config ZSMALLOC_STAT |
| 200 | bool "Export zsmalloc statistics" |
| 201 | depends on ZSMALLOC |
| 202 | select DEBUG_FS |
| 203 | help |
| 204 | This option enables code in the zsmalloc to collect various |
| 205 | statistics about what's happening in zsmalloc and exports that |
| 206 | information to userspace via debugfs. |
| 207 | If unsure, say N. |
| 208 | |
Sergey Senozhatsky | 4ff93b2 | 2023-01-18 09:52:09 +0900 | [diff] [blame] | 209 | config ZSMALLOC_CHAIN_SIZE |
| 210 | int "Maximum number of physical pages per-zspage" |
Sergey Senozhatsky | b46402fa | 2023-01-18 09:52:10 +0900 | [diff] [blame] | 211 | default 8 |
Sergey Senozhatsky | 4ff93b2 | 2023-01-18 09:52:09 +0900 | [diff] [blame] | 212 | range 4 16 |
| 213 | depends on ZSMALLOC |
| 214 | help |
| 215 | This option sets the upper limit on the number of physical pages |
| 216 | that a zmalloc page (zspage) can consist of. The optimal zspage |
| 217 | chain size is calculated for each size class during the |
| 218 | initialization of the pool. |
| 219 | |
| 220 | Changing this option can alter the characteristics of size classes, |
| 221 | such as the number of pages per zspage and the number of objects |
| 222 | per zspage. This can also result in different configurations of |
| 223 | the pool, as zsmalloc merges size classes with similar |
| 224 | characteristics. |
| 225 | |
| 226 | For more information, see zsmalloc documentation. |
| 227 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 228 | menu "SLAB allocator options" |
| 229 | |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 230 | choice |
| 231 | prompt "Choose SLAB allocator" |
| 232 | default SLUB |
| 233 | help |
| 234 | This option allows to select a slab allocator. |
| 235 | |
Vlastimil Babka | eb07c4f39 | 2023-05-23 09:06:34 +0200 | [diff] [blame] | 236 | config SLAB_DEPRECATED |
| 237 | bool "SLAB (DEPRECATED)" |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 238 | depends on !PREEMPT_RT |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 239 | help |
Vlastimil Babka | eb07c4f39 | 2023-05-23 09:06:34 +0200 | [diff] [blame] | 240 | Deprecated and scheduled for removal in a few cycles. Replaced by |
| 241 | SLUB. |
| 242 | |
| 243 | If you cannot migrate to SLUB, please contact linux-mm@kvack.org |
| 244 | and the people listed in the SLAB ALLOCATOR section of MAINTAINERS |
| 245 | file, explaining why. |
| 246 | |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 247 | The regular slab allocator that is established and known to work |
| 248 | well in all environments. It organizes cache hot objects in |
| 249 | per cpu and per node queues. |
| 250 | |
| 251 | config SLUB |
| 252 | bool "SLUB (Unqueued Allocator)" |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 253 | help |
| 254 | SLUB is a slab allocator that minimizes cache line usage |
| 255 | instead of managing queues of cached objects (SLAB approach). |
| 256 | Per cpu caching is realized using slabs of objects instead |
| 257 | of queues of objects. SLUB can use memory efficiently |
| 258 | and has enhanced diagnostics. SLUB is the default choice for |
| 259 | a slab allocator. |
| 260 | |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 261 | endchoice |
| 262 | |
Vlastimil Babka | eb07c4f39 | 2023-05-23 09:06:34 +0200 | [diff] [blame] | 263 | config SLAB |
| 264 | bool |
| 265 | default y |
| 266 | depends on SLAB_DEPRECATED |
| 267 | |
Vlastimil Babka | e240e53 | 2022-11-14 18:13:45 +0100 | [diff] [blame] | 268 | config SLUB_TINY |
| 269 | bool "Configure SLUB for minimal memory footprint" |
| 270 | depends on SLUB && EXPERT |
| 271 | select SLAB_MERGE_DEFAULT |
| 272 | help |
| 273 | Configures the SLUB allocator in a way to achieve minimal memory |
| 274 | footprint, sacrificing scalability, debugging and other features. |
| 275 | This is intended only for the smallest system that had used the |
| 276 | SLOB allocator and is not recommended for systems with more than |
| 277 | 16MB RAM. |
| 278 | |
| 279 | If unsure, say N. |
| 280 | |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 281 | config SLAB_MERGE_DEFAULT |
| 282 | bool "Allow slab caches to be merged" |
| 283 | default y |
| 284 | depends on SLAB || SLUB |
| 285 | help |
| 286 | For reduced kernel memory fragmentation, slab caches can be |
| 287 | merged when they share the same size and other characteristics. |
| 288 | This carries a risk of kernel heap overflows being able to |
| 289 | overwrite objects from merged caches (and more easily control |
| 290 | cache layout), which makes such heap attacks easier to exploit |
| 291 | by attackers. By keeping caches unmerged, these kinds of exploits |
| 292 | can usually only damage objects in the same cache. To disable |
| 293 | merging at runtime, "slab_nomerge" can be passed on the kernel |
| 294 | command line. |
| 295 | |
| 296 | config SLAB_FREELIST_RANDOM |
| 297 | bool "Randomize slab freelist" |
Vlastimil Babka | e240e53 | 2022-11-14 18:13:45 +0100 | [diff] [blame] | 298 | depends on SLAB || (SLUB && !SLUB_TINY) |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 299 | help |
| 300 | Randomizes the freelist order used on creating new pages. This |
| 301 | security feature reduces the predictability of the kernel slab |
| 302 | allocator against heap overflows. |
| 303 | |
| 304 | config SLAB_FREELIST_HARDENED |
| 305 | bool "Harden slab freelist metadata" |
Vlastimil Babka | e240e53 | 2022-11-14 18:13:45 +0100 | [diff] [blame] | 306 | depends on SLAB || (SLUB && !SLUB_TINY) |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 307 | help |
| 308 | Many kernel heap attacks try to target slab cache metadata and |
| 309 | other infrastructure. This options makes minor performance |
| 310 | sacrifices to harden the kernel slab allocator against common |
| 311 | freelist exploit methods. Some slab implementations have more |
| 312 | sanity-checking than others. This option is most effective with |
| 313 | CONFIG_SLUB. |
| 314 | |
Vlastimil Babka | 0710d01 | 2022-05-25 13:25:59 +0200 | [diff] [blame] | 315 | config SLUB_STATS |
| 316 | default n |
| 317 | bool "Enable SLUB performance statistics" |
Vlastimil Babka | e240e53 | 2022-11-14 18:13:45 +0100 | [diff] [blame] | 318 | depends on SLUB && SYSFS && !SLUB_TINY |
Vlastimil Babka | 0710d01 | 2022-05-25 13:25:59 +0200 | [diff] [blame] | 319 | help |
| 320 | SLUB statistics are useful to debug SLUBs allocation behavior in |
| 321 | order find ways to optimize the allocator. This should never be |
| 322 | enabled for production use since keeping statistics slows down |
| 323 | the allocator by a few percentage points. The slabinfo command |
| 324 | supports the determination of the most active slabs to figure |
| 325 | out which slabs are relevant to a particular load. |
| 326 | Try running: slabinfo -DA |
| 327 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 328 | config SLUB_CPU_PARTIAL |
| 329 | default y |
Vlastimil Babka | e240e53 | 2022-11-14 18:13:45 +0100 | [diff] [blame] | 330 | depends on SLUB && SMP && !SLUB_TINY |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 331 | bool "SLUB per cpu partial cache" |
| 332 | help |
| 333 | Per cpu partial caches accelerate objects allocation and freeing |
| 334 | that is local to a processor at the price of more indeterminism |
| 335 | in the latency of the free. On overflow these caches will be cleared |
| 336 | which requires the taking of locks that may cause latency spikes. |
| 337 | Typically one would choose no for a realtime system. |
| 338 | |
GONG, Ruiqi | 3c61529 | 2023-07-14 14:44:22 +0800 | [diff] [blame] | 339 | config RANDOM_KMALLOC_CACHES |
| 340 | default n |
| 341 | depends on SLUB && !SLUB_TINY |
| 342 | bool "Randomize slab caches for normal kmalloc" |
| 343 | help |
| 344 | A hardening feature that creates multiple copies of slab caches for |
| 345 | normal kmalloc allocation and makes kmalloc randomly pick one based |
| 346 | on code address, which makes the attackers more difficult to spray |
| 347 | vulnerable memory objects on the heap for the purpose of exploiting |
| 348 | memory vulnerabilities. |
| 349 | |
| 350 | Currently the number of copies is set to 16, a reasonably large value |
| 351 | that effectively diverges the memory objects allocated for different |
| 352 | subsystems or modules into different caches, at the expense of a |
| 353 | limited degree of memory and CPU overhead that relates to hardware and |
| 354 | system workload. |
| 355 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 356 | endmenu # SLAB allocator options |
| 357 | |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 358 | config SHUFFLE_PAGE_ALLOCATOR |
| 359 | bool "Page allocator randomization" |
| 360 | default SLAB_FREELIST_RANDOM && ACPI_NUMA |
| 361 | help |
| 362 | Randomization of the page allocator improves the average |
| 363 | utilization of a direct-mapped memory-side-cache. See section |
| 364 | 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI |
| 365 | 6.2a specification for an example of how a platform advertises |
| 366 | the presence of a memory-side-cache. There are also incidental |
| 367 | security benefits as it reduces the predictability of page |
| 368 | allocations to compliment SLAB_FREELIST_RANDOM, but the |
Kirill A. Shutemov | 23baf83 | 2023-03-15 14:31:33 +0300 | [diff] [blame] | 369 | default granularity of shuffling on the MAX_ORDER i.e, 10th |
| 370 | order of pages is selected based on cache utilization benefits |
| 371 | on x86. |
Johannes Weiner | 7b42f10 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 372 | |
| 373 | While the randomization improves cache utilization it may |
| 374 | negatively impact workloads on platforms without a cache. For |
| 375 | this reason, by default, the randomization is enabled only |
| 376 | after runtime detection of a direct-mapped memory-side-cache. |
| 377 | Otherwise, the randomization may be force enabled with the |
| 378 | 'page_alloc.shuffle' kernel command line parameter. |
| 379 | |
| 380 | Say Y if unsure. |
| 381 | |
Vlastimil Babka | 0710d01 | 2022-05-25 13:25:59 +0200 | [diff] [blame] | 382 | config COMPAT_BRK |
| 383 | bool "Disable heap randomization" |
| 384 | default y |
| 385 | help |
| 386 | Randomizing heap placement makes heap exploits harder, but it |
| 387 | also breaks ancient binaries (including anything libc5 based). |
| 388 | This option changes the bootup default to heap randomization |
| 389 | disabled, and can be overridden at runtime by setting |
| 390 | /proc/sys/kernel/randomize_va_space to 2. |
| 391 | |
| 392 | On non-ancient distros (post-2000 ones) N is usually a safe choice. |
| 393 | |
| 394 | config MMAP_ALLOW_UNINITIALIZED |
| 395 | bool "Allow mmapped anonymous memory to be uninitialized" |
| 396 | depends on EXPERT && !MMU |
| 397 | default n |
| 398 | help |
| 399 | Normally, and according to the Linux spec, anonymous memory obtained |
| 400 | from mmap() has its contents cleared before it is passed to |
| 401 | userspace. Enabling this config option allows you to request that |
| 402 | mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus |
| 403 | providing a huge performance boost. If this option is not enabled, |
| 404 | then the flag will be ignored. |
| 405 | |
| 406 | This is taken advantage of by uClibc's malloc(), and also by |
| 407 | ELF-FDPIC binfmt's brk and stack allocator. |
| 408 | |
| 409 | Because of the obvious security issues, this option should only be |
| 410 | enabled on embedded devices where you control what is run in |
| 411 | userspace. Since that isn't generally a problem on no-MMU systems, |
| 412 | it is normally safe to say Y here. |
| 413 | |
| 414 | See Documentation/admin-guide/mm/nommu-mmap.rst for more information. |
| 415 | |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 416 | config SELECT_MEMORY_MODEL |
| 417 | def_bool y |
Kees Cook | a8826ee | 2013-01-16 18:54:17 -0800 | [diff] [blame] | 418 | depends on ARCH_SELECT_MEMORY_MODEL |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 419 | |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 420 | choice |
| 421 | prompt "Memory model" |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 422 | depends on SELECT_MEMORY_MODEL |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 423 | default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 424 | default FLATMEM_MANUAL |
Mike Rapoport | d66d109 | 2019-05-13 17:23:05 -0700 | [diff] [blame] | 425 | help |
| 426 | This option allows you to change some of the ways that |
| 427 | Linux manages its memory internally. Most users will |
| 428 | only have one option here selected by the architecture |
| 429 | configuration. This is normal. |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 430 | |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 431 | config FLATMEM_MANUAL |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 432 | bool "Flat Memory" |
Mike Rapoport | bb1c50d | 2021-06-28 19:42:52 -0700 | [diff] [blame] | 433 | depends on !ARCH_SPARSEMEM_ENABLE || ARCH_FLATMEM_ENABLE |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 434 | help |
Mike Rapoport | d66d109 | 2019-05-13 17:23:05 -0700 | [diff] [blame] | 435 | This option is best suited for non-NUMA systems with |
| 436 | flat address space. The FLATMEM is the most efficient |
| 437 | system in terms of performance and resource consumption |
| 438 | and it is the best option for smaller systems. |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 439 | |
Mike Rapoport | d66d109 | 2019-05-13 17:23:05 -0700 | [diff] [blame] | 440 | For systems that have holes in their physical address |
| 441 | spaces and for features like NUMA and memory hotplug, |
Randy Dunlap | dd33d29 | 2019-11-30 17:58:26 -0800 | [diff] [blame] | 442 | choose "Sparse Memory". |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 443 | |
| 444 | If unsure, choose this option (Flat Memory) over any other. |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 445 | |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 446 | config SPARSEMEM_MANUAL |
| 447 | bool "Sparse Memory" |
| 448 | depends on ARCH_SPARSEMEM_ENABLE |
| 449 | help |
| 450 | This will be the only option for some systems, including |
Mike Rapoport | d66d109 | 2019-05-13 17:23:05 -0700 | [diff] [blame] | 451 | memory hot-plug systems. This is normal. |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 452 | |
Mike Rapoport | d66d109 | 2019-05-13 17:23:05 -0700 | [diff] [blame] | 453 | This option provides efficient support for systems with |
| 454 | holes is their physical address space and allows memory |
| 455 | hot-plug and hot-remove. |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 456 | |
Mike Rapoport | d66d109 | 2019-05-13 17:23:05 -0700 | [diff] [blame] | 457 | If unsure, choose "Flat Memory" over this option. |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 458 | |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 459 | endchoice |
| 460 | |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 461 | config SPARSEMEM |
| 462 | def_bool y |
Russell King | 1a83e17 | 2009-10-26 16:50:12 -0700 | [diff] [blame] | 463 | depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 464 | |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 465 | config FLATMEM |
| 466 | def_bool y |
Mike Rapoport | bb1c50d | 2021-06-28 19:42:52 -0700 | [diff] [blame] | 467 | depends on !SPARSEMEM || FLATMEM_MANUAL |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 468 | |
Dave Hansen | 93b7504 | 2005-06-23 00:07:47 -0700 | [diff] [blame] | 469 | # |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 470 | # SPARSEMEM_EXTREME (which is the default) does some bootmem |
Mike Rapoport | c89ab04 | 2020-08-06 23:24:02 -0700 | [diff] [blame] | 471 | # allocations when sparse_init() is called. If this cannot |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 472 | # be done on your architecture, select this option. However, |
| 473 | # statically allocating the mem_section[] array can potentially |
| 474 | # consume vast quantities of .bss, so be careful. |
| 475 | # |
| 476 | # This option will also potentially produce smaller runtime code |
| 477 | # with gcc 3.4 and later. |
| 478 | # |
| 479 | config SPARSEMEM_STATIC |
Jan Beulich | 9ba1608 | 2008-10-15 22:01:38 -0700 | [diff] [blame] | 480 | bool |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 481 | |
| 482 | # |
Matt LaPlante | 44c0920 | 2006-10-03 22:34:14 +0200 | [diff] [blame] | 483 | # Architecture platforms which require a two level mem_section in SPARSEMEM |
Bob Picco | 802f192 | 2005-09-03 15:54:26 -0700 | [diff] [blame] | 484 | # must select this option. This is usually for architecture platforms with |
| 485 | # an extremely sparse physical address space. |
| 486 | # |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 487 | config SPARSEMEM_EXTREME |
| 488 | def_bool y |
| 489 | depends on SPARSEMEM && !SPARSEMEM_STATIC |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 490 | |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 491 | config SPARSEMEM_VMEMMAP_ENABLE |
Jan Beulich | 9ba1608 | 2008-10-15 22:01:38 -0700 | [diff] [blame] | 492 | bool |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 493 | |
| 494 | config SPARSEMEM_VMEMMAP |
Geoff Levand | a5ee6da | 2007-12-17 16:19:53 -0800 | [diff] [blame] | 495 | bool "Sparse Memory virtual memmap" |
| 496 | depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE |
| 497 | default y |
| 498 | help |
Krzysztof Kozlowski | 19fa40a | 2019-11-30 17:58:23 -0800 | [diff] [blame] | 499 | SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise |
| 500 | pfn_to_page and page_to_pfn operations. This is the most |
| 501 | efficient option when sufficient kernel resources are available. |
Aneesh Kumar K.V | 0b376f1 | 2023-04-12 10:30:25 +0530 | [diff] [blame] | 502 | # |
| 503 | # Select this config option from the architecture Kconfig, if it is preferred |
| 504 | # to enable the feature of HugeTLB/dev_dax vmemmap optimization. |
| 505 | # |
Aneesh Kumar K.V | 0b6f158 | 2023-07-25 00:37:53 +0530 | [diff] [blame] | 506 | config ARCH_WANT_OPTIMIZE_DAX_VMEMMAP |
| 507 | bool |
| 508 | |
| 509 | config ARCH_WANT_OPTIMIZE_HUGETLB_VMEMMAP |
Aneesh Kumar K.V | 0b376f1 | 2023-04-12 10:30:25 +0530 | [diff] [blame] | 510 | bool |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 511 | |
Philipp Hachtmann | 70210ed | 2014-01-29 18:16:01 +0100 | [diff] [blame] | 512 | config HAVE_MEMBLOCK_PHYS_MAP |
Christoph Jaeger | 6341e62 | 2014-12-20 15:41:11 -0500 | [diff] [blame] | 513 | bool |
Philipp Hachtmann | 70210ed | 2014-01-29 18:16:01 +0100 | [diff] [blame] | 514 | |
Christoph Hellwig | 67a929e | 2019-07-11 20:57:14 -0700 | [diff] [blame] | 515 | config HAVE_FAST_GUP |
Christoph Hellwig | 050a9ad | 2019-07-11 20:57:21 -0700 | [diff] [blame] | 516 | depends on MMU |
Christoph Jaeger | 6341e62 | 2014-12-20 15:41:11 -0500 | [diff] [blame] | 517 | bool |
Steve Capper | 2667f50 | 2014-10-09 15:29:14 -0700 | [diff] [blame] | 518 | |
David Hildenbrand | 52219ae | 2020-06-04 16:48:38 -0700 | [diff] [blame] | 519 | # Don't discard allocated memory used to track "memory" and "reserved" memblocks |
| 520 | # after early boot, so it can still be used to test for validity of memory. |
| 521 | # Also, memblocks are updated with memory hot(un)plug. |
Mike Rapoport | 350e88b | 2019-05-13 17:22:59 -0700 | [diff] [blame] | 522 | config ARCH_KEEP_MEMBLOCK |
Christoph Jaeger | 6341e62 | 2014-12-20 15:41:11 -0500 | [diff] [blame] | 523 | bool |
Tejun Heo | c378ddd | 2011-07-14 11:46:03 +0200 | [diff] [blame] | 524 | |
Dan Williams | 1e5d8e1 | 2020-02-16 12:01:04 -0800 | [diff] [blame] | 525 | # Keep arch NUMA mapping infrastructure post-init. |
| 526 | config NUMA_KEEP_MEMINFO |
| 527 | bool |
| 528 | |
Minchan Kim | ee6f509 | 2012-07-31 16:43:50 -0700 | [diff] [blame] | 529 | config MEMORY_ISOLATION |
Christoph Jaeger | 6341e62 | 2014-12-20 15:41:11 -0500 | [diff] [blame] | 530 | bool |
Minchan Kim | ee6f509 | 2012-07-31 16:43:50 -0700 | [diff] [blame] | 531 | |
David Hildenbrand | a9e7b8d | 2021-11-08 18:35:50 -0800 | [diff] [blame] | 532 | # IORESOURCE_SYSTEM_RAM regions in the kernel resource tree that are marked |
| 533 | # IORESOURCE_EXCLUSIVE cannot be mapped to user space, for example, via |
| 534 | # /dev/mem. |
| 535 | config EXCLUSIVE_SYSTEM_RAM |
| 536 | def_bool y |
| 537 | depends on !DEVMEM || STRICT_DEVMEM |
| 538 | |
Yasuaki Ishimatsu | 46723bf | 2013-02-22 16:33:00 -0800 | [diff] [blame] | 539 | # |
| 540 | # Only be set on architectures that have completely implemented memory hotplug |
| 541 | # feature. If you are not sure, don't touch it. |
| 542 | # |
| 543 | config HAVE_BOOTMEM_INFO_NODE |
| 544 | def_bool n |
| 545 | |
Anshuman Khandual | 91024b3 | 2021-05-04 18:38:17 -0700 | [diff] [blame] | 546 | config ARCH_ENABLE_MEMORY_HOTPLUG |
| 547 | bool |
| 548 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 549 | config ARCH_ENABLE_MEMORY_HOTREMOVE |
| 550 | bool |
| 551 | |
Dave Hansen | 3947be1 | 2005-10-29 18:16:54 -0700 | [diff] [blame] | 552 | # eventually, we can have this option just 'select SPARSEMEM' |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 553 | menuconfig MEMORY_HOTPLUG |
| 554 | bool "Memory hotplug" |
David Hildenbrand | b30c592 | 2020-10-15 20:08:23 -0700 | [diff] [blame] | 555 | select MEMORY_ISOLATION |
David Hildenbrand | 71b6f2d | 2021-11-05 13:44:20 -0700 | [diff] [blame] | 556 | depends on SPARSEMEM |
Stephen Rothwell | 40b3136 | 2013-05-21 13:49:35 +1000 | [diff] [blame] | 557 | depends on ARCH_ENABLE_MEMORY_HOTPLUG |
David Hildenbrand | 7ec58a2 | 2021-11-05 13:44:28 -0700 | [diff] [blame] | 558 | depends on 64BIT |
Dan Williams | 1e5d8e1 | 2020-02-16 12:01:04 -0800 | [diff] [blame] | 559 | select NUMA_KEEP_MEMINFO if NUMA |
Dave Hansen | 3947be1 | 2005-10-29 18:16:54 -0700 | [diff] [blame] | 560 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 561 | if MEMORY_HOTPLUG |
| 562 | |
Vitaly Kuznetsov | 8604d9e | 2016-05-19 17:13:03 -0700 | [diff] [blame] | 563 | config MEMORY_HOTPLUG_DEFAULT_ONLINE |
Krzysztof Kozlowski | 19fa40a | 2019-11-30 17:58:23 -0800 | [diff] [blame] | 564 | bool "Online the newly added memory blocks by default" |
| 565 | depends on MEMORY_HOTPLUG |
| 566 | help |
Vitaly Kuznetsov | 8604d9e | 2016-05-19 17:13:03 -0700 | [diff] [blame] | 567 | This option sets the default policy setting for memory hotplug |
| 568 | onlining policy (/sys/devices/system/memory/auto_online_blocks) which |
| 569 | determines what happens to newly added memory regions. Policy setting |
| 570 | can always be changed at runtime. |
Mauro Carvalho Chehab | cb1aaeb | 2019-06-07 15:54:32 -0300 | [diff] [blame] | 571 | See Documentation/admin-guide/mm/memory-hotplug.rst for more information. |
Vitaly Kuznetsov | 8604d9e | 2016-05-19 17:13:03 -0700 | [diff] [blame] | 572 | |
| 573 | Say Y here if you want all hot-plugged memory blocks to appear in |
| 574 | 'online' state by default. |
| 575 | Say N here if you want the default policy to keep all hot-plugged |
| 576 | memory blocks in 'offline' state. |
| 577 | |
KAMEZAWA Hiroyuki | 0c0e619 | 2007-10-16 01:26:12 -0700 | [diff] [blame] | 578 | config MEMORY_HOTREMOVE |
| 579 | bool "Allow for memory hot remove" |
Nathan Fontenot | f7e3334 | 2013-09-27 10:18:09 -0500 | [diff] [blame] | 580 | select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64) |
KAMEZAWA Hiroyuki | 0c0e619 | 2007-10-16 01:26:12 -0700 | [diff] [blame] | 581 | depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE |
| 582 | depends on MIGRATION |
| 583 | |
Oscar Salvador | a08a2ae | 2021-05-04 18:39:42 -0700 | [diff] [blame] | 584 | config MHP_MEMMAP_ON_MEMORY |
| 585 | def_bool y |
| 586 | depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP |
| 587 | depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE |
| 588 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 589 | endif # MEMORY_HOTPLUG |
| 590 | |
Aneesh Kumar K.V | 04d5ea4 | 2023-08-08 14:44:56 +0530 | [diff] [blame] | 591 | config ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE |
| 592 | bool |
| 593 | |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 594 | # Heavily threaded applications may benefit from splitting the mm-wide |
| 595 | # page_table_lock, so that faults on different parts of the user address |
| 596 | # space can be handled with less contention: split it at this NR_CPUS. |
| 597 | # Default to 4 for wider testing, though 8 might be more appropriate. |
| 598 | # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock. |
Hugh Dickins | 7b6ac9d | 2005-11-23 13:37:37 -0800 | [diff] [blame] | 599 | # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes. |
Will Deacon | 60bccaa | 2020-05-26 18:33:01 +0100 | [diff] [blame] | 600 | # SPARC32 allocates multiple pte tables within a single page, and therefore |
| 601 | # a per-page lock leads to problems when multiple tables need to be locked |
| 602 | # at the same time (e.g. copy_page_range()). |
Hugh Dickins | a70caa8 | 2009-12-14 17:59:02 -0800 | [diff] [blame] | 603 | # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page. |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 604 | # |
| 605 | config SPLIT_PTLOCK_CPUS |
| 606 | int |
Kirill A. Shutemov | 9164550 | 2014-04-07 15:37:14 -0700 | [diff] [blame] | 607 | default "999999" if !MMU |
Hugh Dickins | a70caa8 | 2009-12-14 17:59:02 -0800 | [diff] [blame] | 608 | default "999999" if ARM && !CPU_CACHE_VIPT |
| 609 | default "999999" if PARISC && !PA20 |
Will Deacon | 60bccaa | 2020-05-26 18:33:01 +0100 | [diff] [blame] | 610 | default "999999" if SPARC32 |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 611 | default "4" |
Christoph Lameter | 7cbe34c | 2006-01-08 01:00:49 -0800 | [diff] [blame] | 612 | |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 613 | config ARCH_ENABLE_SPLIT_PMD_PTLOCK |
Christoph Jaeger | 6341e62 | 2014-12-20 15:41:11 -0500 | [diff] [blame] | 614 | bool |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 615 | |
Christoph Lameter | 7cbe34c | 2006-01-08 01:00:49 -0800 | [diff] [blame] | 616 | # |
Konstantin Khlebnikov | 09316c0 | 2014-10-09 15:29:32 -0700 | [diff] [blame] | 617 | # support for memory balloon |
| 618 | config MEMORY_BALLOON |
Christoph Jaeger | 6341e62 | 2014-12-20 15:41:11 -0500 | [diff] [blame] | 619 | bool |
Konstantin Khlebnikov | 09316c0 | 2014-10-09 15:29:32 -0700 | [diff] [blame] | 620 | |
| 621 | # |
Rafael Aquini | 18468d9 | 2012-12-11 16:02:38 -0800 | [diff] [blame] | 622 | # support for memory balloon compaction |
| 623 | config BALLOON_COMPACTION |
| 624 | bool "Allow for balloon memory compaction/migration" |
| 625 | def_bool y |
Konstantin Khlebnikov | 09316c0 | 2014-10-09 15:29:32 -0700 | [diff] [blame] | 626 | depends on COMPACTION && MEMORY_BALLOON |
Rafael Aquini | 18468d9 | 2012-12-11 16:02:38 -0800 | [diff] [blame] | 627 | help |
| 628 | Memory fragmentation introduced by ballooning might reduce |
| 629 | significantly the number of 2MB contiguous memory blocks that can be |
| 630 | used within a guest, thus imposing performance penalties associated |
| 631 | with the reduced number of transparent huge pages that could be used |
| 632 | by the guest workload. Allowing the compaction & migration for memory |
| 633 | pages enlisted as being part of memory balloon devices avoids the |
| 634 | scenario aforementioned and helps improving memory defragmentation. |
| 635 | |
| 636 | # |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 637 | # support for memory compaction |
| 638 | config COMPACTION |
| 639 | bool "Allow for memory compaction" |
Rik van Riel | 05106e6a | 2012-10-08 16:33:03 -0700 | [diff] [blame] | 640 | def_bool y |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 641 | select MIGRATION |
Andrea Arcangeli | 33a9387 | 2011-01-25 15:07:25 -0800 | [diff] [blame] | 642 | depends on MMU |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 643 | help |
Krzysztof Kozlowski | 19fa40a | 2019-11-30 17:58:23 -0800 | [diff] [blame] | 644 | Compaction is the only memory management component to form |
| 645 | high order (larger physically contiguous) memory blocks |
| 646 | reliably. The page allocator relies on compaction heavily and |
| 647 | the lack of the feature can lead to unexpected OOM killer |
| 648 | invocations for high order memory requests. You shouldn't |
| 649 | disable this option unless there really is a strong reason for |
| 650 | it and then we would be really interested to hear about that at |
| 651 | linux-mm@kvack.org. |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 652 | |
Thomas Gleixner | c7e0b3d | 2022-08-25 18:41:29 +0200 | [diff] [blame] | 653 | config COMPACT_UNEVICTABLE_DEFAULT |
| 654 | int |
| 655 | depends on COMPACTION |
| 656 | default 0 if PREEMPT_RT |
| 657 | default 1 |
| 658 | |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 659 | # |
Alexander Duyck | 36e66c5 | 2020-04-06 20:04:56 -0700 | [diff] [blame] | 660 | # support for free page reporting |
| 661 | config PAGE_REPORTING |
| 662 | bool "Free page reporting" |
| 663 | def_bool n |
| 664 | help |
| 665 | Free page reporting allows for the incremental acquisition of |
| 666 | free pages from the buddy allocator for the purpose of reporting |
| 667 | those pages to another entity, such as a hypervisor, so that the |
| 668 | memory can be freed within the host for other uses. |
| 669 | |
| 670 | # |
Christoph Lameter | 7cbe34c | 2006-01-08 01:00:49 -0800 | [diff] [blame] | 671 | # support for page migration |
| 672 | # |
| 673 | config MIGRATION |
Christoph Lameter | b20a350 | 2006-03-22 00:09:12 -0800 | [diff] [blame] | 674 | bool "Page migration" |
Christoph Lameter | 6c5240a | 2006-06-23 02:03:37 -0700 | [diff] [blame] | 675 | def_bool y |
Chen Gang | de32a81 | 2013-09-12 15:14:08 -0700 | [diff] [blame] | 676 | depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU |
Christoph Lameter | b20a350 | 2006-03-22 00:09:12 -0800 | [diff] [blame] | 677 | help |
| 678 | Allows the migration of the physical location of pages of processes |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 679 | while the virtual addresses are not changed. This is useful in |
| 680 | two situations. The first is on NUMA systems to put pages nearer |
| 681 | to the processors accessing. The second is when allocating huge |
| 682 | pages as migration can relocate pages to satisfy a huge page |
| 683 | allocation instead of reclaiming. |
Greg Kroah-Hartman | 6550e07 | 2006-06-12 17:11:31 -0700 | [diff] [blame] | 684 | |
Christoph Hellwig | 76cbbea | 2022-02-16 15:31:38 +1100 | [diff] [blame] | 685 | config DEVICE_MIGRATION |
Christoph Hellwig | d90a25f | 2022-02-16 15:31:38 +1100 | [diff] [blame] | 686 | def_bool MIGRATION && ZONE_DEVICE |
Christoph Hellwig | 76cbbea | 2022-02-16 15:31:38 +1100 | [diff] [blame] | 687 | |
Naoya Horiguchi | c177c81 | 2014-06-04 16:05:35 -0700 | [diff] [blame] | 688 | config ARCH_ENABLE_HUGEPAGE_MIGRATION |
Christoph Jaeger | 6341e62 | 2014-12-20 15:41:11 -0500 | [diff] [blame] | 689 | bool |
Naoya Horiguchi | c177c81 | 2014-06-04 16:05:35 -0700 | [diff] [blame] | 690 | |
Naoya Horiguchi | 9c670ea | 2017-09-08 16:10:53 -0700 | [diff] [blame] | 691 | config ARCH_ENABLE_THP_MIGRATION |
| 692 | bool |
| 693 | |
Anshuman Khandual | 4bfb68a | 2021-05-04 18:33:19 -0700 | [diff] [blame] | 694 | config HUGETLB_PAGE_SIZE_VARIABLE |
| 695 | def_bool n |
| 696 | help |
| 697 | Allows the pageblock_order value to be dynamic instead of just standard |
| 698 | HUGETLB_PAGE_ORDER when there are multiple HugeTLB page sizes available |
| 699 | on a platform. |
| 700 | |
Kirill A. Shutemov | 23baf83 | 2023-03-15 14:31:33 +0300 | [diff] [blame] | 701 | Note that the pageblock_order cannot exceed MAX_ORDER and will be |
| 702 | clamped down to MAX_ORDER. |
David Hildenbrand | b3d40a2 | 2022-03-22 14:43:20 -0700 | [diff] [blame] | 703 | |
Alexandre Ghiti | 8df995f | 2019-05-13 17:19:00 -0700 | [diff] [blame] | 704 | config CONTIG_ALLOC |
Krzysztof Kozlowski | 19fa40a | 2019-11-30 17:58:23 -0800 | [diff] [blame] | 705 | def_bool (MEMORY_ISOLATION && COMPACTION) || CMA |
Alexandre Ghiti | 8df995f | 2019-05-13 17:19:00 -0700 | [diff] [blame] | 706 | |
Jeremy Fitzhardinge | 600715d | 2008-09-11 01:31:45 -0700 | [diff] [blame] | 707 | config PHYS_ADDR_T_64BIT |
Christoph Hellwig | d4a451d | 2018-04-03 16:24:20 +0200 | [diff] [blame] | 708 | def_bool 64BIT |
Jeremy Fitzhardinge | 600715d | 2008-09-11 01:31:45 -0700 | [diff] [blame] | 709 | |
Christoph Lameter | 2a7326b | 2007-07-17 04:03:37 -0700 | [diff] [blame] | 710 | config BOUNCE |
Vinayak Menon | 9ca24e2 | 2013-04-29 15:08:55 -0700 | [diff] [blame] | 711 | bool "Enable bounce buffers" |
| 712 | default y |
Christoph Hellwig | ce288e0 | 2021-03-31 09:29:59 +0200 | [diff] [blame] | 713 | depends on BLOCK && MMU && HIGHMEM |
Vinayak Menon | 9ca24e2 | 2013-04-29 15:08:55 -0700 | [diff] [blame] | 714 | help |
Christoph Hellwig | ce288e0 | 2021-03-31 09:29:59 +0200 | [diff] [blame] | 715 | Enable bounce buffers for devices that cannot access the full range of |
| 716 | memory available to the CPU. Enabled by default when HIGHMEM is |
| 717 | selected, but you may say n to override this. |
Christoph Lameter | 2a7326b | 2007-07-17 04:03:37 -0700 | [diff] [blame] | 718 | |
Andrea Arcangeli | cddb8a5 | 2008-07-28 15:46:29 -0700 | [diff] [blame] | 719 | config MMU_NOTIFIER |
| 720 | bool |
Jason Gunthorpe | 99cb252 | 2019-11-12 16:22:19 -0400 | [diff] [blame] | 721 | select INTERVAL_TREE |
David Howells | fc4d5c2 | 2009-05-06 16:03:05 -0700 | [diff] [blame] | 722 | |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 723 | config KSM |
| 724 | bool "Enable KSM for page merging" |
| 725 | depends on MMU |
Timofey Titovets | 59e1a2f4 | 2018-12-28 00:34:05 -0800 | [diff] [blame] | 726 | select XXHASH |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 727 | help |
| 728 | Enable Kernel Samepage Merging: KSM periodically scans those areas |
| 729 | of an application's address space that an app has advised may be |
| 730 | mergeable. When it finds pages of identical content, it replaces |
Hugh Dickins | d0f209f | 2009-12-14 17:59:34 -0800 | [diff] [blame] | 731 | the many instances by a single page with that content, so |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 732 | saving memory until one or another app needs to modify the content. |
| 733 | Recommended for use with KVM, or with other duplicative applications. |
Mike Rapoport | ee65728 | 2022-06-27 09:00:26 +0300 | [diff] [blame] | 734 | See Documentation/mm/ksm.rst for more information: KSM is inactive |
Hugh Dickins | c73602a | 2009-10-07 16:32:22 -0700 | [diff] [blame] | 735 | until a program has madvised that an area is MADV_MERGEABLE, and |
| 736 | root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 737 | |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 738 | config DEFAULT_MMAP_MIN_ADDR |
Krzysztof Kozlowski | 19fa40a | 2019-11-30 17:58:23 -0800 | [diff] [blame] | 739 | int "Low address space to protect from user allocation" |
David Howells | 6e14154 | 2009-12-15 19:27:45 +0000 | [diff] [blame] | 740 | depends on MMU |
Krzysztof Kozlowski | 19fa40a | 2019-11-30 17:58:23 -0800 | [diff] [blame] | 741 | default 4096 |
| 742 | help |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 743 | This is the portion of low virtual memory which should be protected |
| 744 | from userspace allocation. Keeping a user from writing to low pages |
| 745 | can help reduce the impact of kernel NULL pointer bugs. |
| 746 | |
| 747 | For most ia64, ppc64 and x86 users with lots of address space |
| 748 | a value of 65536 is reasonable and should cause no problems. |
| 749 | On arm and other archs it should not be higher than 32768. |
Eric Paris | 788084a | 2009-07-31 12:54:11 -0400 | [diff] [blame] | 750 | Programs which use vm86 functionality or have some need to map |
| 751 | this low address space will need CAP_SYS_RAWIO or disable this |
| 752 | protection by setting the value to 0. |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 753 | |
| 754 | This value can be changed after boot using the |
| 755 | /proc/sys/vm/mmap_min_addr tunable. |
| 756 | |
Linus Torvalds | d949f36 | 2009-09-26 09:35:07 -0700 | [diff] [blame] | 757 | config ARCH_SUPPORTS_MEMORY_FAILURE |
| 758 | bool |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 759 | |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 760 | config MEMORY_FAILURE |
| 761 | depends on MMU |
Linus Torvalds | d949f36 | 2009-09-26 09:35:07 -0700 | [diff] [blame] | 762 | depends on ARCH_SUPPORTS_MEMORY_FAILURE |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 763 | bool "Enable recovery from hardware memory errors" |
Minchan Kim | ee6f509 | 2012-07-31 16:43:50 -0700 | [diff] [blame] | 764 | select MEMORY_ISOLATION |
Xie XiuQi | 97f0b13 | 2015-06-24 16:57:36 -0700 | [diff] [blame] | 765 | select RAS |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 766 | help |
| 767 | Enables code to recover from some memory failures on systems |
| 768 | with MCA recovery. This allows a system to continue running |
| 769 | even when some of its memory has uncorrected errors. This requires |
| 770 | special hardware support and typically ECC memory. |
| 771 | |
Andi Kleen | cae681f | 2009-09-16 11:50:17 +0200 | [diff] [blame] | 772 | config HWPOISON_INJECT |
Andi Kleen | 413f9ef | 2009-12-16 12:20:00 +0100 | [diff] [blame] | 773 | tristate "HWPoison pages injector" |
Andi Kleen | 27df506 | 2009-12-21 19:56:42 +0100 | [diff] [blame] | 774 | depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS |
Wu Fengguang | 478c5ff | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 775 | select PROC_PAGE_MONITOR |
Andi Kleen | cae681f | 2009-09-16 11:50:17 +0200 | [diff] [blame] | 776 | |
David Howells | fc4d5c2 | 2009-05-06 16:03:05 -0700 | [diff] [blame] | 777 | config NOMMU_INITIAL_TRIM_EXCESS |
| 778 | int "Turn on mmap() excess space trimming before booting" |
| 779 | depends on !MMU |
| 780 | default 1 |
| 781 | help |
| 782 | The NOMMU mmap() frequently needs to allocate large contiguous chunks |
| 783 | of memory on which to store mappings, but it can only ask the system |
| 784 | allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently |
| 785 | more than it requires. To deal with this, mmap() is able to trim off |
| 786 | the excess and return it to the allocator. |
| 787 | |
| 788 | If trimming is enabled, the excess is trimmed off and returned to the |
| 789 | system allocator, which can cause extra fragmentation, particularly |
| 790 | if there are a lot of transient processes. |
| 791 | |
| 792 | If trimming is disabled, the excess is kept, but not used, which for |
| 793 | long-term mappings means that the space is wasted. |
| 794 | |
| 795 | Trimming can be dynamically controlled through a sysctl option |
| 796 | (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of |
| 797 | excess pages there must be before trimming should occur, or zero if |
| 798 | no trimming is to occur. |
| 799 | |
| 800 | This option specifies the initial value of this option. The default |
| 801 | of 1 says that all excess pages should be trimmed. |
| 802 | |
Stephen Kitt | dd19d29 | 2020-08-12 11:22:30 +0200 | [diff] [blame] | 803 | See Documentation/admin-guide/mm/nommu-mmap.rst for more information. |
Tejun Heo | bbddff0 | 2010-09-03 18:22:48 +0200 | [diff] [blame] | 804 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 805 | config ARCH_WANT_GENERAL_HUGETLB |
| 806 | bool |
| 807 | |
| 808 | config ARCH_WANTS_THP_SWAP |
| 809 | def_bool n |
| 810 | |
| 811 | menuconfig TRANSPARENT_HUGEPAGE |
Andrea Arcangeli | 13ece88 | 2011-01-13 15:47:07 -0800 | [diff] [blame] | 812 | bool "Transparent Hugepage Support" |
Sebastian Andrzej Siewior | 554b0f3 | 2021-11-05 13:35:27 -0700 | [diff] [blame] | 813 | depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE && !PREEMPT_RT |
Andrea Arcangeli | 5d68924 | 2011-01-13 15:47:07 -0800 | [diff] [blame] | 814 | select COMPACTION |
Matthew Wilcox | 3a08cd5 | 2018-09-22 16:14:30 -0400 | [diff] [blame] | 815 | select XARRAY_MULTI |
Andrea Arcangeli | 4c76d9d | 2011-01-13 15:46:39 -0800 | [diff] [blame] | 816 | help |
| 817 | Transparent Hugepages allows the kernel to use huge pages and |
| 818 | huge tlb transparently to the applications whenever possible. |
| 819 | This feature can improve computing performance to certain |
| 820 | applications by speeding up page faults during memory |
| 821 | allocation, by reducing the number of tlb misses and by speeding |
| 822 | up the pagetable walking. |
| 823 | |
| 824 | If memory constrained on embedded, you may want to say N. |
| 825 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 826 | if TRANSPARENT_HUGEPAGE |
| 827 | |
Andrea Arcangeli | 13ece88 | 2011-01-13 15:47:07 -0800 | [diff] [blame] | 828 | choice |
| 829 | prompt "Transparent Hugepage Support sysfs defaults" |
| 830 | depends on TRANSPARENT_HUGEPAGE |
| 831 | default TRANSPARENT_HUGEPAGE_ALWAYS |
| 832 | help |
| 833 | Selects the sysfs defaults for Transparent Hugepage Support. |
| 834 | |
| 835 | config TRANSPARENT_HUGEPAGE_ALWAYS |
| 836 | bool "always" |
| 837 | help |
| 838 | Enabling Transparent Hugepage always, can increase the |
| 839 | memory footprint of applications without a guaranteed |
| 840 | benefit but it will work automatically for all applications. |
| 841 | |
| 842 | config TRANSPARENT_HUGEPAGE_MADVISE |
| 843 | bool "madvise" |
| 844 | help |
| 845 | Enabling Transparent Hugepage madvise, will only provide a |
| 846 | performance improvement benefit to the applications using |
| 847 | madvise(MADV_HUGEPAGE) but it won't risk to increase the |
| 848 | memory footprint of applications without a guaranteed |
| 849 | benefit. |
| 850 | endchoice |
| 851 | |
Huang Ying | 38d8b4e | 2017-07-06 15:37:18 -0700 | [diff] [blame] | 852 | config THP_SWAP |
| 853 | def_bool y |
Hugh Dickins | dad6a5e | 2022-11-02 18:48:45 -0700 | [diff] [blame] | 854 | depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP && 64BIT |
Huang Ying | 38d8b4e | 2017-07-06 15:37:18 -0700 | [diff] [blame] | 855 | help |
| 856 | Swap transparent huge pages in one piece, without splitting. |
Huang Ying | 14fef28 | 2018-08-17 15:49:41 -0700 | [diff] [blame] | 857 | XXX: For now, swap cluster backing transparent huge page |
| 858 | will be split after swapout. |
Huang Ying | 38d8b4e | 2017-07-06 15:37:18 -0700 | [diff] [blame] | 859 | |
| 860 | For selection by architectures with reasonable THP sizes. |
| 861 | |
Johannes Weiner | 519bcb79 | 2022-05-19 14:08:53 -0700 | [diff] [blame] | 862 | config READ_ONLY_THP_FOR_FS |
| 863 | bool "Read-only THP for filesystems (EXPERIMENTAL)" |
| 864 | depends on TRANSPARENT_HUGEPAGE && SHMEM |
| 865 | |
| 866 | help |
| 867 | Allow khugepaged to put read-only file-backed pages in THP. |
| 868 | |
| 869 | This is marked experimental because it is a new feature. Write |
| 870 | support of file THPs will be developed in the next few release |
| 871 | cycles. |
| 872 | |
| 873 | endif # TRANSPARENT_HUGEPAGE |
| 874 | |
Kirill A. Shutemov | e496cf3 | 2016-07-26 15:26:35 -0700 | [diff] [blame] | 875 | # |
Tejun Heo | bbddff0 | 2010-09-03 18:22:48 +0200 | [diff] [blame] | 876 | # UP and nommu archs use km based percpu allocator |
| 877 | # |
| 878 | config NEED_PER_CPU_KM |
Vladimir Murzin | 3583521 | 2021-11-30 17:29:54 +0000 | [diff] [blame] | 879 | depends on !SMP || !MMU |
Tejun Heo | bbddff0 | 2010-09-03 18:22:48 +0200 | [diff] [blame] | 880 | bool |
| 881 | default y |
Dan Magenheimer | 077b1f8 | 2011-05-26 10:01:36 -0600 | [diff] [blame] | 882 | |
Kefeng Wang | 7ecd19c | 2022-01-19 18:07:41 -0800 | [diff] [blame] | 883 | config NEED_PER_CPU_EMBED_FIRST_CHUNK |
| 884 | bool |
| 885 | |
| 886 | config NEED_PER_CPU_PAGE_FIRST_CHUNK |
| 887 | bool |
| 888 | |
| 889 | config USE_PERCPU_NUMA_NODE_ID |
| 890 | bool |
| 891 | |
| 892 | config HAVE_SETUP_PER_CPU_AREA |
| 893 | bool |
| 894 | |
Aneesh Kumar K.V | f825c73 | 2013-07-02 11:15:15 +0530 | [diff] [blame] | 895 | config CMA |
| 896 | bool "Contiguous Memory Allocator" |
Mike Rapoport | aca52c3 | 2018-10-30 15:07:44 -0700 | [diff] [blame] | 897 | depends on MMU |
Aneesh Kumar K.V | f825c73 | 2013-07-02 11:15:15 +0530 | [diff] [blame] | 898 | select MIGRATION |
| 899 | select MEMORY_ISOLATION |
| 900 | help |
| 901 | This enables the Contiguous Memory Allocator which allows other |
| 902 | subsystems to allocate big physically-contiguous blocks of memory. |
| 903 | CMA reserves a region of memory and allows only movable pages to |
| 904 | be allocated from it. This way, the kernel can use the memory for |
| 905 | pagecache and when a subsystem requests for contiguous area, the |
| 906 | allocated pages are migrated away to serve the contiguous request. |
| 907 | |
| 908 | If unsure, say "n". |
| 909 | |
| 910 | config CMA_DEBUG |
| 911 | bool "CMA debug messages (DEVELOPMENT)" |
| 912 | depends on DEBUG_KERNEL && CMA |
| 913 | help |
| 914 | Turns on debug messages in CMA. This produces KERN_DEBUG |
| 915 | messages for every CMA call as well as various messages while |
| 916 | processing calls such as dma_alloc_from_contiguous(). |
| 917 | This option does not affect warning and error messages. |
Alexander Graf | bf550fc | 2013-08-29 00:41:59 +0200 | [diff] [blame] | 918 | |
Sasha Levin | 28b24c1 | 2015-04-14 15:44:57 -0700 | [diff] [blame] | 919 | config CMA_DEBUGFS |
| 920 | bool "CMA debugfs interface" |
| 921 | depends on CMA && DEBUG_FS |
| 922 | help |
| 923 | Turns on the DebugFS interface for CMA. |
| 924 | |
Minchan Kim | 43ca106 | 2021-05-04 18:37:28 -0700 | [diff] [blame] | 925 | config CMA_SYSFS |
| 926 | bool "CMA information through sysfs interface" |
| 927 | depends on CMA && SYSFS |
| 928 | help |
| 929 | This option exposes some sysfs attributes to get information |
| 930 | from CMA. |
| 931 | |
Joonsoo Kim | a254129 | 2014-08-06 16:05:25 -0700 | [diff] [blame] | 932 | config CMA_AREAS |
| 933 | int "Maximum count of the CMA areas" |
| 934 | depends on CMA |
Barry Song | b7176c2 | 2020-08-24 11:03:07 +1200 | [diff] [blame] | 935 | default 19 if NUMA |
Joonsoo Kim | a254129 | 2014-08-06 16:05:25 -0700 | [diff] [blame] | 936 | default 7 |
| 937 | help |
| 938 | CMA allows to create CMA areas for particular purpose, mainly, |
| 939 | used as device private area. This parameter sets the maximum |
| 940 | number of CMA area in the system. |
| 941 | |
Barry Song | b7176c2 | 2020-08-24 11:03:07 +1200 | [diff] [blame] | 942 | If unsure, leave the default value "7" in UMA and "19" in NUMA. |
Joonsoo Kim | a254129 | 2014-08-06 16:05:25 -0700 | [diff] [blame] | 943 | |
Dan Streetman | af8d417 | 2014-08-06 16:08:36 -0700 | [diff] [blame] | 944 | config MEM_SOFT_DIRTY |
| 945 | bool "Track memory changes" |
| 946 | depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS |
| 947 | select PROC_PAGE_MONITOR |
Seth Jennings | 4e2e277 | 2013-07-10 16:04:55 -0700 | [diff] [blame] | 948 | help |
Dan Streetman | af8d417 | 2014-08-06 16:08:36 -0700 | [diff] [blame] | 949 | This option enables memory changes tracking by introducing a |
| 950 | soft-dirty bit on pte-s. This bit it set when someone writes |
| 951 | into a page just as regular dirty bit, but unlike the latter |
| 952 | it can be cleared by hands. |
| 953 | |
Mike Rapoport | 1ad1335 | 2018-04-18 11:07:49 +0300 | [diff] [blame] | 954 | See Documentation/admin-guide/mm/soft-dirty.rst for more details. |
Seth Jennings | 4e2e277 | 2013-07-10 16:04:55 -0700 | [diff] [blame] | 955 | |
Mark Salter | 9e5c33d | 2014-04-07 15:39:48 -0700 | [diff] [blame] | 956 | config GENERIC_EARLY_IOREMAP |
| 957 | bool |
Helge Deller | 042d27a | 2014-04-30 23:26:02 +0200 | [diff] [blame] | 958 | |
Helge Deller | 22ee3ea | 2020-11-06 19:41:36 +0100 | [diff] [blame] | 959 | config STACK_MAX_DEFAULT_SIZE_MB |
| 960 | int "Default maximum user stack size for 32-bit processes (MB)" |
| 961 | default 100 |
Helge Deller | 042d27a | 2014-04-30 23:26:02 +0200 | [diff] [blame] | 962 | range 8 2048 |
| 963 | depends on STACK_GROWSUP && (!64BIT || COMPAT) |
| 964 | help |
| 965 | This is the maximum stack size in Megabytes in the VM layout of 32-bit |
| 966 | user processes when the stack grows upwards (currently only on parisc |
Helge Deller | 22ee3ea | 2020-11-06 19:41:36 +0100 | [diff] [blame] | 967 | arch) when the RLIMIT_STACK hard limit is unlimited. |
Helge Deller | 042d27a | 2014-04-30 23:26:02 +0200 | [diff] [blame] | 968 | |
Helge Deller | 22ee3ea | 2020-11-06 19:41:36 +0100 | [diff] [blame] | 969 | A sane initial value is 100 MB. |
Mel Gorman | 3a80a7f | 2015-06-30 14:57:02 -0700 | [diff] [blame] | 970 | |
Mel Gorman | 3a80a7f | 2015-06-30 14:57:02 -0700 | [diff] [blame] | 971 | config DEFERRED_STRUCT_PAGE_INIT |
Vlastimil Babka | 1ce2210 | 2016-02-05 15:36:21 -0800 | [diff] [blame] | 972 | bool "Defer initialisation of struct pages to kthreads" |
Mike Rapoport | d39f8fb | 2018-08-17 15:47:07 -0700 | [diff] [blame] | 973 | depends on SPARSEMEM |
Pavel Tatashin | ab1e8d8 | 2018-05-18 16:09:13 -0700 | [diff] [blame] | 974 | depends on !NEED_PER_CPU_KM |
Pasha Tatashin | 889c695 | 2018-09-20 12:22:30 -0700 | [diff] [blame] | 975 | depends on 64BIT |
Daniel Jordan | e444314 | 2020-06-03 15:59:51 -0700 | [diff] [blame] | 976 | select PADATA |
Mel Gorman | 3a80a7f | 2015-06-30 14:57:02 -0700 | [diff] [blame] | 977 | help |
| 978 | Ordinarily all struct pages are initialised during early boot in a |
| 979 | single thread. On very large machines this can take a considerable |
| 980 | amount of time. If this option is set, large machines will bring up |
Daniel Jordan | e444314 | 2020-06-03 15:59:51 -0700 | [diff] [blame] | 981 | a subset of memmap at boot and then initialise the rest in parallel. |
| 982 | This has a potential performance impact on tasks running early in the |
Vlastimil Babka | 1ce2210 | 2016-02-05 15:36:21 -0800 | [diff] [blame] | 983 | lifetime of the system until these kthreads finish the |
| 984 | initialisation. |
Dan Williams | 033fbae | 2015-08-09 15:29:06 -0400 | [diff] [blame] | 985 | |
SeongJae Park | 1c676e0 | 2021-09-07 19:56:40 -0700 | [diff] [blame] | 986 | config PAGE_IDLE_FLAG |
| 987 | bool |
| 988 | select PAGE_EXTENSION if !64BIT |
| 989 | help |
| 990 | This adds PG_idle and PG_young flags to 'struct page'. PTE Accessed |
| 991 | bit writers can set the state of the bit in the flags so that PTE |
| 992 | Accessed bit readers may avoid disturbance. |
| 993 | |
Vladimir Davydov | 33c3fc7 | 2015-09-09 15:35:45 -0700 | [diff] [blame] | 994 | config IDLE_PAGE_TRACKING |
| 995 | bool "Enable idle page tracking" |
| 996 | depends on SYSFS && MMU |
SeongJae Park | 1c676e0 | 2021-09-07 19:56:40 -0700 | [diff] [blame] | 997 | select PAGE_IDLE_FLAG |
Vladimir Davydov | 33c3fc7 | 2015-09-09 15:35:45 -0700 | [diff] [blame] | 998 | help |
| 999 | This feature allows to estimate the amount of user pages that have |
| 1000 | not been touched during a given period of time. This information can |
| 1001 | be useful to tune memory cgroup limits and/or for job placement |
| 1002 | within a compute cluster. |
| 1003 | |
Mike Rapoport | 1ad1335 | 2018-04-18 11:07:49 +0300 | [diff] [blame] | 1004 | See Documentation/admin-guide/mm/idle_page_tracking.rst for |
| 1005 | more details. |
Vladimir Davydov | 33c3fc7 | 2015-09-09 15:35:45 -0700 | [diff] [blame] | 1006 | |
Anshuman Khandual | c2280be | 2021-05-04 18:38:09 -0700 | [diff] [blame] | 1007 | config ARCH_HAS_CACHE_LINE_SIZE |
| 1008 | bool |
| 1009 | |
Kees Cook | 2792d84 | 2022-02-16 12:05:28 -0800 | [diff] [blame] | 1010 | config ARCH_HAS_CURRENT_STACK_POINTER |
| 1011 | bool |
| 1012 | help |
| 1013 | In support of HARDENED_USERCOPY performing stack variable lifetime |
| 1014 | checking, an architecture-agnostic way to find the stack pointer |
| 1015 | is needed. Once an architecture defines an unsigned long global |
| 1016 | register alias named "current_stack_pointer", this config can be |
| 1017 | selected. |
| 1018 | |
Robin Murphy | 1759673 | 2019-07-16 16:30:47 -0700 | [diff] [blame] | 1019 | config ARCH_HAS_PTE_DEVMAP |
Oliver O'Halloran | 65f7d04 | 2017-06-28 11:32:31 +1000 | [diff] [blame] | 1020 | bool |
| 1021 | |
Kefeng Wang | 63703f3 | 2021-06-30 18:52:20 -0700 | [diff] [blame] | 1022 | config ARCH_HAS_ZONE_DMA_SET |
| 1023 | bool |
| 1024 | |
| 1025 | config ZONE_DMA |
| 1026 | bool "Support DMA zone" if ARCH_HAS_ZONE_DMA_SET |
| 1027 | default y if ARM64 || X86 |
| 1028 | |
| 1029 | config ZONE_DMA32 |
| 1030 | bool "Support DMA32 zone" if ARCH_HAS_ZONE_DMA_SET |
| 1031 | depends on !X86_32 |
| 1032 | default y if ARM64 |
| 1033 | |
Dan Williams | 033fbae | 2015-08-09 15:29:06 -0400 | [diff] [blame] | 1034 | config ZONE_DEVICE |
Jérôme Glisse | 5042db4 | 2017-09-08 16:11:43 -0700 | [diff] [blame] | 1035 | bool "Device memory (pmem, HMM, etc...) hotplug support" |
Dan Williams | 033fbae | 2015-08-09 15:29:06 -0400 | [diff] [blame] | 1036 | depends on MEMORY_HOTPLUG |
| 1037 | depends on MEMORY_HOTREMOVE |
Dan Williams | 99490f1 | 2016-03-17 14:19:58 -0700 | [diff] [blame] | 1038 | depends on SPARSEMEM_VMEMMAP |
Robin Murphy | 1759673 | 2019-07-16 16:30:47 -0700 | [diff] [blame] | 1039 | depends on ARCH_HAS_PTE_DEVMAP |
Matthew Wilcox | 3a08cd5 | 2018-09-22 16:14:30 -0400 | [diff] [blame] | 1040 | select XARRAY_MULTI |
Dan Williams | 033fbae | 2015-08-09 15:29:06 -0400 | [diff] [blame] | 1041 | |
| 1042 | help |
| 1043 | Device memory hotplug support allows for establishing pmem, |
| 1044 | or other device driver discovered memory regions, in the |
| 1045 | memmap. This allows pfn_to_page() lookups of otherwise |
| 1046 | "device-physical" addresses which is needed for using a DAX |
| 1047 | mapping in an O_DIRECT operation, among other things. |
| 1048 | |
| 1049 | If FS_DAX is enabled, then say Y. |
Linus Torvalds | 06a660a | 2015-09-11 16:42:39 -0700 | [diff] [blame] | 1050 | |
Christoph Hellwig | 9c240a7 | 2019-08-06 19:05:52 +0300 | [diff] [blame] | 1051 | # |
| 1052 | # Helpers to mirror range of the CPU page tables of a process into device page |
| 1053 | # tables. |
| 1054 | # |
Jérôme Glisse | c0b1240 | 2017-09-08 16:11:27 -0700 | [diff] [blame] | 1055 | config HMM_MIRROR |
Christoph Hellwig | 9c240a7 | 2019-08-06 19:05:52 +0300 | [diff] [blame] | 1056 | bool |
Christoph Hellwig | f442c28 | 2019-08-06 19:05:51 +0300 | [diff] [blame] | 1057 | depends on MMU |
Jérôme Glisse | c0b1240 | 2017-09-08 16:11:27 -0700 | [diff] [blame] | 1058 | |
Dan Williams | 14b80582 | 2022-05-20 13:41:24 -0700 | [diff] [blame] | 1059 | config GET_FREE_REGION |
| 1060 | depends on SPARSEMEM |
| 1061 | bool |
| 1062 | |
Jérôme Glisse | 5042db4 | 2017-09-08 16:11:43 -0700 | [diff] [blame] | 1063 | config DEVICE_PRIVATE |
| 1064 | bool "Unaddressable device memory (GPU memory, ...)" |
Christoph Hellwig | 7328d9c | 2019-06-26 14:27:22 +0200 | [diff] [blame] | 1065 | depends on ZONE_DEVICE |
Dan Williams | 14b80582 | 2022-05-20 13:41:24 -0700 | [diff] [blame] | 1066 | select GET_FREE_REGION |
Jérôme Glisse | 5042db4 | 2017-09-08 16:11:43 -0700 | [diff] [blame] | 1067 | |
| 1068 | help |
| 1069 | Allows creation of struct pages to represent unaddressable device |
| 1070 | memory; i.e., memory that is only accessible from the device (or |
| 1071 | group of devices). You likely also want to select HMM_MIRROR. |
| 1072 | |
Christoph Hellwig | 3e9a9e2 | 2020-10-17 16:15:10 -0700 | [diff] [blame] | 1073 | config VMAP_PFN |
| 1074 | bool |
| 1075 | |
Dave Hansen | 63c17fb | 2016-02-12 13:02:08 -0800 | [diff] [blame] | 1076 | config ARCH_USES_HIGH_VMA_FLAGS |
| 1077 | bool |
Dave Hansen | 66d3757 | 2016-02-12 13:02:32 -0800 | [diff] [blame] | 1078 | config ARCH_HAS_PKEYS |
| 1079 | bool |
Dennis Zhou | 30a5b53 | 2017-06-19 19:28:31 -0400 | [diff] [blame] | 1080 | |
Catalin Marinas | b0284cd | 2022-11-03 18:10:34 -0700 | [diff] [blame] | 1081 | config ARCH_USES_PG_ARCH_X |
| 1082 | bool |
| 1083 | help |
| 1084 | Enable the definition of PG_arch_x page flags with x > 1. Only |
| 1085 | suitable for 64-bit architectures with CONFIG_FLATMEM or |
| 1086 | CONFIG_SPARSEMEM_VMEMMAP enabled, otherwise there may not be |
| 1087 | enough room for additional bits in page->flags. |
| 1088 | |
Vlastimil Babka | 0710d01 | 2022-05-25 13:25:59 +0200 | [diff] [blame] | 1089 | config VM_EVENT_COUNTERS |
| 1090 | default y |
| 1091 | bool "Enable VM event counters for /proc/vmstat" if EXPERT |
| 1092 | help |
| 1093 | VM event counters are needed for event counts to be shown. |
| 1094 | This option allows the disabling of the VM event counters |
| 1095 | on EXPERT systems. /proc/vmstat will only show page counts |
| 1096 | if VM event counters are disabled. |
| 1097 | |
Dennis Zhou | 30a5b53 | 2017-06-19 19:28:31 -0400 | [diff] [blame] | 1098 | config PERCPU_STATS |
| 1099 | bool "Collect percpu memory statistics" |
Dennis Zhou | 30a5b53 | 2017-06-19 19:28:31 -0400 | [diff] [blame] | 1100 | help |
| 1101 | This feature collects and exposes statistics via debugfs. The |
| 1102 | information includes global and per chunk statistics, which can |
| 1103 | be used to help understand percpu memory usage. |
Kirill A. Shutemov | 64c349f4 | 2017-11-17 15:31:22 -0800 | [diff] [blame] | 1104 | |
John Hubbard | 9c84f22 | 2020-12-14 19:05:05 -0800 | [diff] [blame] | 1105 | config GUP_TEST |
| 1106 | bool "Enable infrastructure for get_user_pages()-related unit tests" |
Barry Song | d0de824 | 2020-12-14 19:05:38 -0800 | [diff] [blame] | 1107 | depends on DEBUG_FS |
Kirill A. Shutemov | 64c349f4 | 2017-11-17 15:31:22 -0800 | [diff] [blame] | 1108 | help |
John Hubbard | 9c84f22 | 2020-12-14 19:05:05 -0800 | [diff] [blame] | 1109 | Provides /sys/kernel/debug/gup_test, which in turn provides a way |
| 1110 | to make ioctl calls that can launch kernel-based unit tests for |
| 1111 | the get_user_pages*() and pin_user_pages*() family of API calls. |
Kirill A. Shutemov | 64c349f4 | 2017-11-17 15:31:22 -0800 | [diff] [blame] | 1112 | |
John Hubbard | 9c84f22 | 2020-12-14 19:05:05 -0800 | [diff] [blame] | 1113 | These tests include benchmark testing of the _fast variants of |
| 1114 | get_user_pages*() and pin_user_pages*(), as well as smoke tests of |
| 1115 | the non-_fast variants. |
| 1116 | |
John Hubbard | f4f9bda | 2020-12-14 19:05:21 -0800 | [diff] [blame] | 1117 | There is also a sub-test that allows running dump_page() on any |
| 1118 | of up to eight pages (selected by command line args) within the |
| 1119 | range of user-space addresses. These pages are either pinned via |
| 1120 | pin_user_pages*(), or pinned via get_user_pages*(), as specified |
| 1121 | by other command line arguments. |
| 1122 | |
SeongJae Park | baa489f | 2023-01-03 18:07:53 +0000 | [diff] [blame] | 1123 | See tools/testing/selftests/mm/gup_test.c |
Laurent Dufour | 3010a5e | 2018-06-07 17:06:08 -0700 | [diff] [blame] | 1124 | |
Barry Song | d0de824 | 2020-12-14 19:05:38 -0800 | [diff] [blame] | 1125 | comment "GUP_TEST needs to have DEBUG_FS enabled" |
| 1126 | depends on !GUP_TEST && !DEBUG_FS |
Laurent Dufour | 3010a5e | 2018-06-07 17:06:08 -0700 | [diff] [blame] | 1127 | |
Peter Zijlstra | 6ca297d | 2022-10-21 14:51:44 +0200 | [diff] [blame] | 1128 | config GUP_GET_PXX_LOW_HIGH |
Christoph Hellwig | 39656e8 | 2019-07-11 20:56:49 -0700 | [diff] [blame] | 1129 | bool |
| 1130 | |
Keith Busch | def8574 | 2023-01-26 13:51:14 -0800 | [diff] [blame] | 1131 | config DMAPOOL_TEST |
| 1132 | tristate "Enable a module to run time tests on dma_pool" |
| 1133 | depends on HAS_DMA |
| 1134 | help |
| 1135 | Provides a test module that will allocate and free many blocks of |
| 1136 | various sizes and report how long it takes. This is intended to |
| 1137 | provide a consistent way to measure how changes to the |
| 1138 | dma_pool_alloc/free routines affect performance. |
| 1139 | |
Laurent Dufour | 3010a5e | 2018-06-07 17:06:08 -0700 | [diff] [blame] | 1140 | config ARCH_HAS_PTE_SPECIAL |
| 1141 | bool |
Christoph Hellwig | 59e0b52 | 2018-07-31 13:39:35 +0200 | [diff] [blame] | 1142 | |
Christoph Hellwig | cbd34da | 2019-07-11 20:57:28 -0700 | [diff] [blame] | 1143 | # |
| 1144 | # Some architectures require a special hugepage directory format that is |
| 1145 | # required to support multiple hugepage sizes. For example a4fe3ce76 |
| 1146 | # "powerpc/mm: Allow more flexible layouts for hugepage pagetables" |
| 1147 | # introduced it on powerpc. This allows for a more flexible hugepage |
| 1148 | # pagetable layouts. |
| 1149 | # |
| 1150 | config ARCH_HAS_HUGEPD |
| 1151 | bool |
| 1152 | |
Thomas Hellstrom | c5acad8 | 2019-03-19 13:12:30 +0100 | [diff] [blame] | 1153 | config MAPPING_DIRTY_HELPERS |
| 1154 | bool |
| 1155 | |
Thomas Gleixner | 298fa1a | 2020-11-03 10:27:18 +0100 | [diff] [blame] | 1156 | config KMAP_LOCAL |
| 1157 | bool |
| 1158 | |
Ard Biesheuvel | 825c43f | 2021-11-19 16:43:55 -0800 | [diff] [blame] | 1159 | config KMAP_LOCAL_NON_LINEAR_PTE_ARRAY |
| 1160 | bool |
| 1161 | |
Christoph Hellwig | 1fbaf8f | 2021-04-29 22:57:32 -0700 | [diff] [blame] | 1162 | # struct io_mapping based helper. Selected by drivers that need them |
| 1163 | config IO_MAPPING |
| 1164 | bool |
Mike Rapoport | 1507f51 | 2021-07-07 18:08:03 -0700 | [diff] [blame] | 1165 | |
Thomas Weißschuh | 626e98c | 2023-06-30 11:08:53 +0200 | [diff] [blame] | 1166 | config MEMFD_CREATE |
| 1167 | bool "Enable memfd_create() system call" if EXPERT |
| 1168 | |
Mike Rapoport | 1507f51 | 2021-07-07 18:08:03 -0700 | [diff] [blame] | 1169 | config SECRETMEM |
Lukas Bulwahn | 7494772 | 2022-11-16 14:19:22 +0100 | [diff] [blame] | 1170 | default y |
| 1171 | bool "Enable memfd_secret() system call" if EXPERT |
| 1172 | depends on ARCH_HAS_SET_DIRECT_MAP |
| 1173 | help |
| 1174 | Enable the memfd_secret() system call with the ability to create |
| 1175 | memory areas visible only in the context of the owning process and |
| 1176 | not mapped to other processes and other kernel page tables. |
Mike Rapoport | 1507f51 | 2021-07-07 18:08:03 -0700 | [diff] [blame] | 1177 | |
Colin Cross | 9a10064 | 2022-01-14 14:05:59 -0800 | [diff] [blame] | 1178 | config ANON_VMA_NAME |
| 1179 | bool "Anonymous VMA name support" |
| 1180 | depends on PROC_FS && ADVISE_SYSCALLS && MMU |
| 1181 | |
| 1182 | help |
| 1183 | Allow naming anonymous virtual memory areas. |
| 1184 | |
| 1185 | This feature allows assigning names to virtual memory areas. Assigned |
| 1186 | names can be later retrieved from /proc/pid/maps and /proc/pid/smaps |
| 1187 | and help identifying individual anonymous memory areas. |
| 1188 | Assigning a name to anonymous virtual memory area might prevent that |
| 1189 | area from being merged with adjacent virtual memory areas due to the |
| 1190 | difference in their name. |
| 1191 | |
Peter Xu | 430529b | 2022-05-12 20:22:56 -0700 | [diff] [blame] | 1192 | config USERFAULTFD |
| 1193 | bool "Enable userfaultfd() system call" |
| 1194 | depends on MMU |
| 1195 | help |
| 1196 | Enable the userfaultfd() system call that allows to intercept and |
| 1197 | handle page faults in userland. |
| 1198 | |
| 1199 | config HAVE_ARCH_USERFAULTFD_WP |
| 1200 | bool |
| 1201 | help |
| 1202 | Arch has userfaultfd write protection support |
| 1203 | |
| 1204 | config HAVE_ARCH_USERFAULTFD_MINOR |
| 1205 | bool |
| 1206 | help |
| 1207 | Arch has userfaultfd minor fault support |
| 1208 | |
Peter Xu | 1db9dbc | 2022-05-12 20:22:52 -0700 | [diff] [blame] | 1209 | config PTE_MARKER_UFFD_WP |
Peter Xu | 81e0f15 | 2022-05-12 20:22:56 -0700 | [diff] [blame] | 1210 | bool "Userfaultfd write protection support for shmem/hugetlbfs" |
| 1211 | default y |
| 1212 | depends on HAVE_ARCH_USERFAULTFD_WP |
Peter Xu | 1db9dbc | 2022-05-12 20:22:52 -0700 | [diff] [blame] | 1213 | |
| 1214 | help |
| 1215 | Allows to create marker PTEs for userfaultfd write protection |
| 1216 | purposes. It is required to enable userfaultfd write protection on |
| 1217 | file-backed memory types like shmem and hugetlbfs. |
| 1218 | |
Yu Zhao | ac35a49 | 2022-09-18 02:00:03 -0600 | [diff] [blame] | 1219 | # multi-gen LRU { |
Yu Zhao | ec1c86b | 2022-09-18 02:00:02 -0600 | [diff] [blame] | 1220 | config LRU_GEN |
| 1221 | bool "Multi-Gen LRU" |
| 1222 | depends on MMU |
| 1223 | # make sure folio->flags has enough spare bits |
| 1224 | depends on 64BIT || !SPARSEMEM || SPARSEMEM_VMEMMAP |
| 1225 | help |
Yu Zhao | 07017ac | 2022-09-18 02:00:10 -0600 | [diff] [blame] | 1226 | A high performance LRU implementation to overcommit memory. See |
| 1227 | Documentation/admin-guide/mm/multigen_lru.rst for details. |
Yu Zhao | ec1c86b | 2022-09-18 02:00:02 -0600 | [diff] [blame] | 1228 | |
Yu Zhao | 354ed59 | 2022-09-18 02:00:07 -0600 | [diff] [blame] | 1229 | config LRU_GEN_ENABLED |
| 1230 | bool "Enable by default" |
| 1231 | depends on LRU_GEN |
| 1232 | help |
| 1233 | This option enables the multi-gen LRU by default. |
| 1234 | |
Yu Zhao | ac35a49 | 2022-09-18 02:00:03 -0600 | [diff] [blame] | 1235 | config LRU_GEN_STATS |
| 1236 | bool "Full stats for debugging" |
| 1237 | depends on LRU_GEN |
| 1238 | help |
| 1239 | Do not enable this option unless you plan to look at historical stats |
| 1240 | from evicted generations for debugging purpose. |
| 1241 | |
| 1242 | This option has a per-memcg and per-node memory overhead. |
| 1243 | # } |
| 1244 | |
Suren Baghdasaryan | 0b6cc04 | 2023-02-27 09:36:08 -0800 | [diff] [blame] | 1245 | config ARCH_SUPPORTS_PER_VMA_LOCK |
| 1246 | def_bool n |
| 1247 | |
| 1248 | config PER_VMA_LOCK |
| 1249 | def_bool y |
| 1250 | depends on ARCH_SUPPORTS_PER_VMA_LOCK && MMU && SMP |
| 1251 | help |
| 1252 | Allow per-vma locking during page fault handling. |
| 1253 | |
| 1254 | This feature allows locking each virtual memory area separately when |
| 1255 | handling page faults instead of taking mmap_lock. |
| 1256 | |
Linus Torvalds | c2508ec | 2023-06-15 15:17:36 -0700 | [diff] [blame] | 1257 | config LOCK_MM_AND_FIND_VMA |
| 1258 | bool |
| 1259 | depends on !STACK_GROWSUP |
| 1260 | |
SeongJae Park | 2224d84 | 2021-09-07 19:56:28 -0700 | [diff] [blame] | 1261 | source "mm/damon/Kconfig" |
| 1262 | |
Christoph Hellwig | 59e0b52 | 2018-07-31 13:39:35 +0200 | [diff] [blame] | 1263 | endmenu |