]>
Commit | Line | Data |
---|---|---|
ec8f24b7 | 1 | # SPDX-License-Identifier: GPL-2.0-only |
59e0b520 CH |
2 | |
3 | menu "Memory Management options" | |
4 | ||
e1785e85 DH |
5 | config SELECT_MEMORY_MODEL |
6 | def_bool y | |
a8826eeb | 7 | depends on ARCH_SELECT_MEMORY_MODEL |
e1785e85 | 8 | |
3a9da765 DH |
9 | choice |
10 | prompt "Memory model" | |
e1785e85 | 11 | depends on SELECT_MEMORY_MODEL |
d41dee36 | 12 | default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT |
e1785e85 | 13 | default FLATMEM_MANUAL |
d66d109d MR |
14 | help |
15 | This option allows you to change some of the ways that | |
16 | Linux manages its memory internally. Most users will | |
17 | only have one option here selected by the architecture | |
18 | configuration. This is normal. | |
3a9da765 | 19 | |
e1785e85 | 20 | config FLATMEM_MANUAL |
3a9da765 | 21 | bool "Flat Memory" |
c898ec16 | 22 | depends on !(ARCH_DISCONTIGMEM_ENABLE || ARCH_SPARSEMEM_ENABLE) || ARCH_FLATMEM_ENABLE |
3a9da765 | 23 | help |
d66d109d MR |
24 | This option is best suited for non-NUMA systems with |
25 | flat address space. The FLATMEM is the most efficient | |
26 | system in terms of performance and resource consumption | |
27 | and it is the best option for smaller systems. | |
28 | ||
29 | For systems that have holes in their physical address | |
30 | spaces and for features like NUMA and memory hotplug, | |
dd33d29a | 31 | choose "Sparse Memory". |
d41dee36 AW |
32 | |
33 | If unsure, choose this option (Flat Memory) over any other. | |
3a9da765 | 34 | |
e1785e85 | 35 | config DISCONTIGMEM_MANUAL |
f3519f91 | 36 | bool "Discontiguous Memory" |
3a9da765 DH |
37 | depends on ARCH_DISCONTIGMEM_ENABLE |
38 | help | |
785dcd44 DH |
39 | This option provides enhanced support for discontiguous |
40 | memory systems, over FLATMEM. These systems have holes | |
41 | in their physical address spaces, and this option provides | |
d66d109d | 42 | more efficient handling of these holes. |
785dcd44 | 43 | |
d66d109d MR |
44 | Although "Discontiguous Memory" is still used by several |
45 | architectures, it is considered deprecated in favor of | |
46 | "Sparse Memory". | |
785dcd44 | 47 | |
d66d109d | 48 | If unsure, choose "Sparse Memory" over this option. |
3a9da765 | 49 | |
d41dee36 AW |
50 | config SPARSEMEM_MANUAL |
51 | bool "Sparse Memory" | |
52 | depends on ARCH_SPARSEMEM_ENABLE | |
53 | help | |
54 | This will be the only option for some systems, including | |
d66d109d | 55 | memory hot-plug systems. This is normal. |
d41dee36 | 56 | |
d66d109d MR |
57 | This option provides efficient support for systems with |
58 | holes is their physical address space and allows memory | |
59 | hot-plug and hot-remove. | |
d41dee36 | 60 | |
d66d109d | 61 | If unsure, choose "Flat Memory" over this option. |
d41dee36 | 62 | |
3a9da765 DH |
63 | endchoice |
64 | ||
e1785e85 DH |
65 | config DISCONTIGMEM |
66 | def_bool y | |
67 | depends on (!SELECT_MEMORY_MODEL && ARCH_DISCONTIGMEM_ENABLE) || DISCONTIGMEM_MANUAL | |
68 | ||
d41dee36 AW |
69 | config SPARSEMEM |
70 | def_bool y | |
1a83e175 | 71 | depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL |
d41dee36 | 72 | |
e1785e85 DH |
73 | config FLATMEM |
74 | def_bool y | |
d41dee36 AW |
75 | depends on (!DISCONTIGMEM && !SPARSEMEM) || FLATMEM_MANUAL |
76 | ||
77 | config FLAT_NODE_MEM_MAP | |
78 | def_bool y | |
79 | depends on !SPARSEMEM | |
e1785e85 | 80 | |
93b7504e DH |
81 | # |
82 | # Both the NUMA code and DISCONTIGMEM use arrays of pg_data_t's | |
83 | # to represent different areas of memory. This variable allows | |
84 | # those dependencies to exist individually. | |
85 | # | |
86 | config NEED_MULTIPLE_NODES | |
87 | def_bool y | |
88 | depends on DISCONTIGMEM || NUMA | |
af705362 | 89 | |
3e347261 BP |
90 | # |
91 | # SPARSEMEM_EXTREME (which is the default) does some bootmem | |
c89ab04f | 92 | # allocations when sparse_init() is called. If this cannot |
3e347261 BP |
93 | # be done on your architecture, select this option. However, |
94 | # statically allocating the mem_section[] array can potentially | |
95 | # consume vast quantities of .bss, so be careful. | |
96 | # | |
97 | # This option will also potentially produce smaller runtime code | |
98 | # with gcc 3.4 and later. | |
99 | # | |
100 | config SPARSEMEM_STATIC | |
9ba16087 | 101 | bool |
3e347261 | 102 | |
802f192e | 103 | # |
44c09201 | 104 | # Architecture platforms which require a two level mem_section in SPARSEMEM |
802f192e BP |
105 | # must select this option. This is usually for architecture platforms with |
106 | # an extremely sparse physical address space. | |
107 | # | |
3e347261 BP |
108 | config SPARSEMEM_EXTREME |
109 | def_bool y | |
110 | depends on SPARSEMEM && !SPARSEMEM_STATIC | |
4c21e2f2 | 111 | |
29c71111 | 112 | config SPARSEMEM_VMEMMAP_ENABLE |
9ba16087 | 113 | bool |
29c71111 AW |
114 | |
115 | config SPARSEMEM_VMEMMAP | |
a5ee6daa GL |
116 | bool "Sparse Memory virtual memmap" |
117 | depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE | |
118 | default y | |
119 | help | |
19fa40a0 KK |
120 | SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise |
121 | pfn_to_page and page_to_pfn operations. This is the most | |
122 | efficient option when sufficient kernel resources are available. | |
29c71111 | 123 | |
70210ed9 | 124 | config HAVE_MEMBLOCK_PHYS_MAP |
6341e62b | 125 | bool |
70210ed9 | 126 | |
67a929e0 | 127 | config HAVE_FAST_GUP |
050a9adc | 128 | depends on MMU |
6341e62b | 129 | bool |
2667f50e | 130 | |
52219aea DH |
131 | # Don't discard allocated memory used to track "memory" and "reserved" memblocks |
132 | # after early boot, so it can still be used to test for validity of memory. | |
133 | # Also, memblocks are updated with memory hot(un)plug. | |
350e88ba | 134 | config ARCH_KEEP_MEMBLOCK |
6341e62b | 135 | bool |
c378ddd5 | 136 | |
1e5d8e1e DW |
137 | # Keep arch NUMA mapping infrastructure post-init. |
138 | config NUMA_KEEP_MEMINFO | |
139 | bool | |
140 | ||
ee6f509c | 141 | config MEMORY_ISOLATION |
6341e62b | 142 | bool |
ee6f509c | 143 | |
46723bfa YI |
144 | # |
145 | # Only be set on architectures that have completely implemented memory hotplug | |
146 | # feature. If you are not sure, don't touch it. | |
147 | # | |
148 | config HAVE_BOOTMEM_INFO_NODE | |
149 | def_bool n | |
150 | ||
91024b3c AK |
151 | config ARCH_ENABLE_MEMORY_HOTPLUG |
152 | bool | |
153 | ||
3947be19 DH |
154 | # eventually, we can have this option just 'select SPARSEMEM' |
155 | config MEMORY_HOTPLUG | |
156 | bool "Allow for memory hot-add" | |
b30c5927 | 157 | select MEMORY_ISOLATION |
ec69acbb | 158 | depends on SPARSEMEM || X86_64_ACPI_NUMA |
40b31360 | 159 | depends on ARCH_ENABLE_MEMORY_HOTPLUG |
b59d02ed | 160 | depends on 64BIT || BROKEN |
1e5d8e1e | 161 | select NUMA_KEEP_MEMINFO if NUMA |
3947be19 | 162 | |
ec69acbb KM |
163 | config MEMORY_HOTPLUG_SPARSE |
164 | def_bool y | |
165 | depends on SPARSEMEM && MEMORY_HOTPLUG | |
166 | ||
8604d9e5 | 167 | config MEMORY_HOTPLUG_DEFAULT_ONLINE |
19fa40a0 KK |
168 | bool "Online the newly added memory blocks by default" |
169 | depends on MEMORY_HOTPLUG | |
170 | help | |
8604d9e5 VK |
171 | This option sets the default policy setting for memory hotplug |
172 | onlining policy (/sys/devices/system/memory/auto_online_blocks) which | |
173 | determines what happens to newly added memory regions. Policy setting | |
174 | can always be changed at runtime. | |
cb1aaebe | 175 | See Documentation/admin-guide/mm/memory-hotplug.rst for more information. |
8604d9e5 VK |
176 | |
177 | Say Y here if you want all hot-plugged memory blocks to appear in | |
178 | 'online' state by default. | |
179 | Say N here if you want the default policy to keep all hot-plugged | |
180 | memory blocks in 'offline' state. | |
181 | ||
91024b3c AK |
182 | config ARCH_ENABLE_MEMORY_HOTREMOVE |
183 | bool | |
184 | ||
0c0e6195 KH |
185 | config MEMORY_HOTREMOVE |
186 | bool "Allow for memory hot remove" | |
f7e3334a | 187 | select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64) |
0c0e6195 KH |
188 | depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE |
189 | depends on MIGRATION | |
190 | ||
a08a2ae3 OS |
191 | config MHP_MEMMAP_ON_MEMORY |
192 | def_bool y | |
193 | depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP | |
194 | depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE | |
195 | ||
4c21e2f2 HD |
196 | # Heavily threaded applications may benefit from splitting the mm-wide |
197 | # page_table_lock, so that faults on different parts of the user address | |
198 | # space can be handled with less contention: split it at this NR_CPUS. | |
199 | # Default to 4 for wider testing, though 8 might be more appropriate. | |
200 | # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock. | |
7b6ac9df | 201 | # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes. |
60bccaa6 WD |
202 | # SPARC32 allocates multiple pte tables within a single page, and therefore |
203 | # a per-page lock leads to problems when multiple tables need to be locked | |
204 | # at the same time (e.g. copy_page_range()). | |
a70caa8b | 205 | # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page. |
4c21e2f2 HD |
206 | # |
207 | config SPLIT_PTLOCK_CPUS | |
208 | int | |
9164550e | 209 | default "999999" if !MMU |
a70caa8b HD |
210 | default "999999" if ARM && !CPU_CACHE_VIPT |
211 | default "999999" if PARISC && !PA20 | |
60bccaa6 | 212 | default "999999" if SPARC32 |
4c21e2f2 | 213 | default "4" |
7cbe34cf | 214 | |
e009bb30 | 215 | config ARCH_ENABLE_SPLIT_PMD_PTLOCK |
6341e62b | 216 | bool |
e009bb30 | 217 | |
09316c09 KK |
218 | # |
219 | # support for memory balloon | |
220 | config MEMORY_BALLOON | |
6341e62b | 221 | bool |
09316c09 | 222 | |
18468d93 RA |
223 | # |
224 | # support for memory balloon compaction | |
225 | config BALLOON_COMPACTION | |
226 | bool "Allow for balloon memory compaction/migration" | |
227 | def_bool y | |
09316c09 | 228 | depends on COMPACTION && MEMORY_BALLOON |
18468d93 RA |
229 | help |
230 | Memory fragmentation introduced by ballooning might reduce | |
231 | significantly the number of 2MB contiguous memory blocks that can be | |
232 | used within a guest, thus imposing performance penalties associated | |
233 | with the reduced number of transparent huge pages that could be used | |
234 | by the guest workload. Allowing the compaction & migration for memory | |
235 | pages enlisted as being part of memory balloon devices avoids the | |
236 | scenario aforementioned and helps improving memory defragmentation. | |
237 | ||
e9e96b39 MG |
238 | # |
239 | # support for memory compaction | |
240 | config COMPACTION | |
241 | bool "Allow for memory compaction" | |
05106e6a | 242 | def_bool y |
e9e96b39 | 243 | select MIGRATION |
33a93877 | 244 | depends on MMU |
e9e96b39 | 245 | help |
19fa40a0 KK |
246 | Compaction is the only memory management component to form |
247 | high order (larger physically contiguous) memory blocks | |
248 | reliably. The page allocator relies on compaction heavily and | |
249 | the lack of the feature can lead to unexpected OOM killer | |
250 | invocations for high order memory requests. You shouldn't | |
251 | disable this option unless there really is a strong reason for | |
252 | it and then we would be really interested to hear about that at | |
253 | [email protected]. | |
e9e96b39 | 254 | |
36e66c55 AD |
255 | # |
256 | # support for free page reporting | |
257 | config PAGE_REPORTING | |
258 | bool "Free page reporting" | |
259 | def_bool n | |
260 | help | |
261 | Free page reporting allows for the incremental acquisition of | |
262 | free pages from the buddy allocator for the purpose of reporting | |
263 | those pages to another entity, such as a hypervisor, so that the | |
264 | memory can be freed within the host for other uses. | |
265 | ||
7cbe34cf CL |
266 | # |
267 | # support for page migration | |
268 | # | |
269 | config MIGRATION | |
b20a3503 | 270 | bool "Page migration" |
6c5240ae | 271 | def_bool y |
de32a817 | 272 | depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU |
b20a3503 CL |
273 | help |
274 | Allows the migration of the physical location of pages of processes | |
e9e96b39 MG |
275 | while the virtual addresses are not changed. This is useful in |
276 | two situations. The first is on NUMA systems to put pages nearer | |
277 | to the processors accessing. The second is when allocating huge | |
278 | pages as migration can relocate pages to satisfy a huge page | |
279 | allocation instead of reclaiming. | |
6550e07f | 280 | |
c177c81e | 281 | config ARCH_ENABLE_HUGEPAGE_MIGRATION |
6341e62b | 282 | bool |
c177c81e | 283 | |
9c670ea3 NH |
284 | config ARCH_ENABLE_THP_MIGRATION |
285 | bool | |
286 | ||
4bfb68a0 AK |
287 | config HUGETLB_PAGE_SIZE_VARIABLE |
288 | def_bool n | |
289 | help | |
290 | Allows the pageblock_order value to be dynamic instead of just standard | |
291 | HUGETLB_PAGE_ORDER when there are multiple HugeTLB page sizes available | |
292 | on a platform. | |
293 | ||
8df995f6 | 294 | config CONTIG_ALLOC |
19fa40a0 | 295 | def_bool (MEMORY_ISOLATION && COMPACTION) || CMA |
8df995f6 | 296 | |
600715dc | 297 | config PHYS_ADDR_T_64BIT |
d4a451d5 | 298 | def_bool 64BIT |
600715dc | 299 | |
2a7326b5 | 300 | config BOUNCE |
9ca24e2e VM |
301 | bool "Enable bounce buffers" |
302 | default y | |
ce288e05 | 303 | depends on BLOCK && MMU && HIGHMEM |
9ca24e2e | 304 | help |
ce288e05 CH |
305 | Enable bounce buffers for devices that cannot access the full range of |
306 | memory available to the CPU. Enabled by default when HIGHMEM is | |
307 | selected, but you may say n to override this. | |
2a7326b5 | 308 | |
f057eac0 | 309 | config VIRT_TO_BUS |
4febd95a SR |
310 | bool |
311 | help | |
312 | An architecture should select this if it implements the | |
313 | deprecated interface virt_to_bus(). All new architectures | |
314 | should probably not select this. | |
315 | ||
cddb8a5c AA |
316 | |
317 | config MMU_NOTIFIER | |
318 | bool | |
83fe27ea | 319 | select SRCU |
99cb252f | 320 | select INTERVAL_TREE |
fc4d5c29 | 321 | |
f8af4da3 HD |
322 | config KSM |
323 | bool "Enable KSM for page merging" | |
324 | depends on MMU | |
59e1a2f4 | 325 | select XXHASH |
f8af4da3 HD |
326 | help |
327 | Enable Kernel Samepage Merging: KSM periodically scans those areas | |
328 | of an application's address space that an app has advised may be | |
329 | mergeable. When it finds pages of identical content, it replaces | |
d0f209f6 | 330 | the many instances by a single page with that content, so |
f8af4da3 HD |
331 | saving memory until one or another app needs to modify the content. |
332 | Recommended for use with KVM, or with other duplicative applications. | |
ad56b738 | 333 | See Documentation/vm/ksm.rst for more information: KSM is inactive |
c73602ad HD |
334 | until a program has madvised that an area is MADV_MERGEABLE, and |
335 | root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). | |
f8af4da3 | 336 | |
e0a94c2a | 337 | config DEFAULT_MMAP_MIN_ADDR |
19fa40a0 | 338 | int "Low address space to protect from user allocation" |
6e141546 | 339 | depends on MMU |
19fa40a0 KK |
340 | default 4096 |
341 | help | |
e0a94c2a CL |
342 | This is the portion of low virtual memory which should be protected |
343 | from userspace allocation. Keeping a user from writing to low pages | |
344 | can help reduce the impact of kernel NULL pointer bugs. | |
345 | ||
346 | For most ia64, ppc64 and x86 users with lots of address space | |
347 | a value of 65536 is reasonable and should cause no problems. | |
348 | On arm and other archs it should not be higher than 32768. | |
788084ab EP |
349 | Programs which use vm86 functionality or have some need to map |
350 | this low address space will need CAP_SYS_RAWIO or disable this | |
351 | protection by setting the value to 0. | |
e0a94c2a CL |
352 | |
353 | This value can be changed after boot using the | |
354 | /proc/sys/vm/mmap_min_addr tunable. | |
355 | ||
d949f36f LT |
356 | config ARCH_SUPPORTS_MEMORY_FAILURE |
357 | bool | |
e0a94c2a | 358 | |
6a46079c AK |
359 | config MEMORY_FAILURE |
360 | depends on MMU | |
d949f36f | 361 | depends on ARCH_SUPPORTS_MEMORY_FAILURE |
6a46079c | 362 | bool "Enable recovery from hardware memory errors" |
ee6f509c | 363 | select MEMORY_ISOLATION |
97f0b134 | 364 | select RAS |
6a46079c AK |
365 | help |
366 | Enables code to recover from some memory failures on systems | |
367 | with MCA recovery. This allows a system to continue running | |
368 | even when some of its memory has uncorrected errors. This requires | |
369 | special hardware support and typically ECC memory. | |
370 | ||
cae681fc | 371 | config HWPOISON_INJECT |
413f9efb | 372 | tristate "HWPoison pages injector" |
27df5068 | 373 | depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS |
478c5ffc | 374 | select PROC_PAGE_MONITOR |
cae681fc | 375 | |
fc4d5c29 DH |
376 | config NOMMU_INITIAL_TRIM_EXCESS |
377 | int "Turn on mmap() excess space trimming before booting" | |
378 | depends on !MMU | |
379 | default 1 | |
380 | help | |
381 | The NOMMU mmap() frequently needs to allocate large contiguous chunks | |
382 | of memory on which to store mappings, but it can only ask the system | |
383 | allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently | |
384 | more than it requires. To deal with this, mmap() is able to trim off | |
385 | the excess and return it to the allocator. | |
386 | ||
387 | If trimming is enabled, the excess is trimmed off and returned to the | |
388 | system allocator, which can cause extra fragmentation, particularly | |
389 | if there are a lot of transient processes. | |
390 | ||
391 | If trimming is disabled, the excess is kept, but not used, which for | |
392 | long-term mappings means that the space is wasted. | |
393 | ||
394 | Trimming can be dynamically controlled through a sysctl option | |
395 | (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of | |
396 | excess pages there must be before trimming should occur, or zero if | |
397 | no trimming is to occur. | |
398 | ||
399 | This option specifies the initial value of this option. The default | |
400 | of 1 says that all excess pages should be trimmed. | |
401 | ||
dd19d293 | 402 | See Documentation/admin-guide/mm/nommu-mmap.rst for more information. |
bbddff05 | 403 | |
4c76d9d1 | 404 | config TRANSPARENT_HUGEPAGE |
13ece886 | 405 | bool "Transparent Hugepage Support" |
15626062 | 406 | depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE |
5d689240 | 407 | select COMPACTION |
3a08cd52 | 408 | select XARRAY_MULTI |
4c76d9d1 AA |
409 | help |
410 | Transparent Hugepages allows the kernel to use huge pages and | |
411 | huge tlb transparently to the applications whenever possible. | |
412 | This feature can improve computing performance to certain | |
413 | applications by speeding up page faults during memory | |
414 | allocation, by reducing the number of tlb misses and by speeding | |
415 | up the pagetable walking. | |
416 | ||
417 | If memory constrained on embedded, you may want to say N. | |
418 | ||
13ece886 AA |
419 | choice |
420 | prompt "Transparent Hugepage Support sysfs defaults" | |
421 | depends on TRANSPARENT_HUGEPAGE | |
422 | default TRANSPARENT_HUGEPAGE_ALWAYS | |
423 | help | |
424 | Selects the sysfs defaults for Transparent Hugepage Support. | |
425 | ||
426 | config TRANSPARENT_HUGEPAGE_ALWAYS | |
427 | bool "always" | |
428 | help | |
429 | Enabling Transparent Hugepage always, can increase the | |
430 | memory footprint of applications without a guaranteed | |
431 | benefit but it will work automatically for all applications. | |
432 | ||
433 | config TRANSPARENT_HUGEPAGE_MADVISE | |
434 | bool "madvise" | |
435 | help | |
436 | Enabling Transparent Hugepage madvise, will only provide a | |
437 | performance improvement benefit to the applications using | |
438 | madvise(MADV_HUGEPAGE) but it won't risk to increase the | |
439 | memory footprint of applications without a guaranteed | |
440 | benefit. | |
441 | endchoice | |
442 | ||
38d8b4e6 | 443 | config ARCH_WANTS_THP_SWAP |
19fa40a0 | 444 | def_bool n |
38d8b4e6 YH |
445 | |
446 | config THP_SWAP | |
447 | def_bool y | |
14fef284 | 448 | depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP |
38d8b4e6 YH |
449 | help |
450 | Swap transparent huge pages in one piece, without splitting. | |
14fef284 YH |
451 | XXX: For now, swap cluster backing transparent huge page |
452 | will be split after swapout. | |
38d8b4e6 YH |
453 | |
454 | For selection by architectures with reasonable THP sizes. | |
455 | ||
bbddff05 TH |
456 | # |
457 | # UP and nommu archs use km based percpu allocator | |
458 | # | |
459 | config NEED_PER_CPU_KM | |
460 | depends on !SMP | |
461 | bool | |
462 | default y | |
077b1f83 DM |
463 | |
464 | config CLEANCACHE | |
465 | bool "Enable cleancache driver to cache clean pages if tmem is present" | |
077b1f83 DM |
466 | help |
467 | Cleancache can be thought of as a page-granularity victim cache | |
468 | for clean pages that the kernel's pageframe replacement algorithm | |
469 | (PFRA) would like to keep around, but can't since there isn't enough | |
470 | memory. So when the PFRA "evicts" a page, it first attempts to use | |
140a1ef2 | 471 | cleancache code to put the data contained in that page into |
077b1f83 DM |
472 | "transcendent memory", memory that is not directly accessible or |
473 | addressable by the kernel and is of unknown and possibly | |
474 | time-varying size. And when a cleancache-enabled | |
475 | filesystem wishes to access a page in a file on disk, it first | |
476 | checks cleancache to see if it already contains it; if it does, | |
477 | the page is copied into the kernel and a disk access is avoided. | |
478 | When a transcendent memory driver is available (such as zcache or | |
479 | Xen transcendent memory), a significant I/O reduction | |
480 | may be achieved. When none is available, all cleancache calls | |
481 | are reduced to a single pointer-compare-against-NULL resulting | |
482 | in a negligible performance hit. | |
483 | ||
484 | If unsure, say Y to enable cleancache | |
27c6aec2 DM |
485 | |
486 | config FRONTSWAP | |
487 | bool "Enable frontswap to cache swap pages if tmem is present" | |
488 | depends on SWAP | |
27c6aec2 DM |
489 | help |
490 | Frontswap is so named because it can be thought of as the opposite | |
491 | of a "backing" store for a swap device. The data is stored into | |
492 | "transcendent memory", memory that is not directly accessible or | |
493 | addressable by the kernel and is of unknown and possibly | |
494 | time-varying size. When space in transcendent memory is available, | |
495 | a significant swap I/O reduction may be achieved. When none is | |
496 | available, all frontswap calls are reduced to a single pointer- | |
497 | compare-against-NULL resulting in a negligible performance hit | |
498 | and swap data is stored as normal on the matching swap device. | |
499 | ||
500 | If unsure, say Y to enable frontswap. | |
f825c736 AK |
501 | |
502 | config CMA | |
503 | bool "Contiguous Memory Allocator" | |
aca52c39 | 504 | depends on MMU |
f825c736 AK |
505 | select MIGRATION |
506 | select MEMORY_ISOLATION | |
507 | help | |
508 | This enables the Contiguous Memory Allocator which allows other | |
509 | subsystems to allocate big physically-contiguous blocks of memory. | |
510 | CMA reserves a region of memory and allows only movable pages to | |
511 | be allocated from it. This way, the kernel can use the memory for | |
512 | pagecache and when a subsystem requests for contiguous area, the | |
513 | allocated pages are migrated away to serve the contiguous request. | |
514 | ||
515 | If unsure, say "n". | |
516 | ||
517 | config CMA_DEBUG | |
518 | bool "CMA debug messages (DEVELOPMENT)" | |
519 | depends on DEBUG_KERNEL && CMA | |
520 | help | |
521 | Turns on debug messages in CMA. This produces KERN_DEBUG | |
522 | messages for every CMA call as well as various messages while | |
523 | processing calls such as dma_alloc_from_contiguous(). | |
524 | This option does not affect warning and error messages. | |
bf550fc9 | 525 | |
28b24c1f SL |
526 | config CMA_DEBUGFS |
527 | bool "CMA debugfs interface" | |
528 | depends on CMA && DEBUG_FS | |
529 | help | |
530 | Turns on the DebugFS interface for CMA. | |
531 | ||
43ca106f MK |
532 | config CMA_SYSFS |
533 | bool "CMA information through sysfs interface" | |
534 | depends on CMA && SYSFS | |
535 | help | |
536 | This option exposes some sysfs attributes to get information | |
537 | from CMA. | |
538 | ||
a254129e JK |
539 | config CMA_AREAS |
540 | int "Maximum count of the CMA areas" | |
541 | depends on CMA | |
b7176c26 | 542 | default 19 if NUMA |
a254129e JK |
543 | default 7 |
544 | help | |
545 | CMA allows to create CMA areas for particular purpose, mainly, | |
546 | used as device private area. This parameter sets the maximum | |
547 | number of CMA area in the system. | |
548 | ||
b7176c26 | 549 | If unsure, leave the default value "7" in UMA and "19" in NUMA. |
a254129e | 550 | |
af8d417a DS |
551 | config MEM_SOFT_DIRTY |
552 | bool "Track memory changes" | |
553 | depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS | |
554 | select PROC_PAGE_MONITOR | |
4e2e2770 | 555 | help |
af8d417a DS |
556 | This option enables memory changes tracking by introducing a |
557 | soft-dirty bit on pte-s. This bit it set when someone writes | |
558 | into a page just as regular dirty bit, but unlike the latter | |
559 | it can be cleared by hands. | |
560 | ||
1ad1335d | 561 | See Documentation/admin-guide/mm/soft-dirty.rst for more details. |
4e2e2770 | 562 | |
2b281117 SJ |
563 | config ZSWAP |
564 | bool "Compressed cache for swap pages (EXPERIMENTAL)" | |
565 | depends on FRONTSWAP && CRYPTO=y | |
12d79d64 | 566 | select ZPOOL |
2b281117 SJ |
567 | help |
568 | A lightweight compressed cache for swap pages. It takes | |
569 | pages that are in the process of being swapped out and attempts to | |
570 | compress them into a dynamically allocated RAM-based memory pool. | |
571 | This can result in a significant I/O reduction on swap device and, | |
572 | in the case where decompressing from RAM is faster that swap device | |
573 | reads, can also improve workload performance. | |
574 | ||
575 | This is marked experimental because it is a new feature (as of | |
576 | v3.11) that interacts heavily with memory reclaim. While these | |
577 | interactions don't cause any known issues on simple memory setups, | |
578 | they have not be fully explored on the large set of potential | |
579 | configurations and workloads that exist. | |
580 | ||
bb8b93b5 MS |
581 | choice |
582 | prompt "Compressed cache for swap pages default compressor" | |
583 | depends on ZSWAP | |
584 | default ZSWAP_COMPRESSOR_DEFAULT_LZO | |
585 | help | |
586 | Selects the default compression algorithm for the compressed cache | |
587 | for swap pages. | |
588 | ||
589 | For an overview what kind of performance can be expected from | |
590 | a particular compression algorithm please refer to the benchmarks | |
591 | available at the following LWN page: | |
592 | https://lwn.net/Articles/751795/ | |
593 | ||
594 | If in doubt, select 'LZO'. | |
595 | ||
596 | The selection made here can be overridden by using the kernel | |
597 | command line 'zswap.compressor=' option. | |
598 | ||
599 | config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE | |
600 | bool "Deflate" | |
601 | select CRYPTO_DEFLATE | |
602 | help | |
603 | Use the Deflate algorithm as the default compression algorithm. | |
604 | ||
605 | config ZSWAP_COMPRESSOR_DEFAULT_LZO | |
606 | bool "LZO" | |
607 | select CRYPTO_LZO | |
608 | help | |
609 | Use the LZO algorithm as the default compression algorithm. | |
610 | ||
611 | config ZSWAP_COMPRESSOR_DEFAULT_842 | |
612 | bool "842" | |
613 | select CRYPTO_842 | |
614 | help | |
615 | Use the 842 algorithm as the default compression algorithm. | |
616 | ||
617 | config ZSWAP_COMPRESSOR_DEFAULT_LZ4 | |
618 | bool "LZ4" | |
619 | select CRYPTO_LZ4 | |
620 | help | |
621 | Use the LZ4 algorithm as the default compression algorithm. | |
622 | ||
623 | config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC | |
624 | bool "LZ4HC" | |
625 | select CRYPTO_LZ4HC | |
626 | help | |
627 | Use the LZ4HC algorithm as the default compression algorithm. | |
628 | ||
629 | config ZSWAP_COMPRESSOR_DEFAULT_ZSTD | |
630 | bool "zstd" | |
631 | select CRYPTO_ZSTD | |
632 | help | |
633 | Use the zstd algorithm as the default compression algorithm. | |
634 | endchoice | |
635 | ||
636 | config ZSWAP_COMPRESSOR_DEFAULT | |
637 | string | |
638 | depends on ZSWAP | |
639 | default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE | |
640 | default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO | |
641 | default "842" if ZSWAP_COMPRESSOR_DEFAULT_842 | |
642 | default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4 | |
643 | default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC | |
644 | default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD | |
645 | default "" | |
646 | ||
647 | choice | |
648 | prompt "Compressed cache for swap pages default allocator" | |
649 | depends on ZSWAP | |
650 | default ZSWAP_ZPOOL_DEFAULT_ZBUD | |
651 | help | |
652 | Selects the default allocator for the compressed cache for | |
653 | swap pages. | |
654 | The default is 'zbud' for compatibility, however please do | |
655 | read the description of each of the allocators below before | |
656 | making a right choice. | |
657 | ||
658 | The selection made here can be overridden by using the kernel | |
659 | command line 'zswap.zpool=' option. | |
660 | ||
661 | config ZSWAP_ZPOOL_DEFAULT_ZBUD | |
662 | bool "zbud" | |
663 | select ZBUD | |
664 | help | |
665 | Use the zbud allocator as the default allocator. | |
666 | ||
667 | config ZSWAP_ZPOOL_DEFAULT_Z3FOLD | |
668 | bool "z3fold" | |
669 | select Z3FOLD | |
670 | help | |
671 | Use the z3fold allocator as the default allocator. | |
672 | ||
673 | config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC | |
674 | bool "zsmalloc" | |
675 | select ZSMALLOC | |
676 | help | |
677 | Use the zsmalloc allocator as the default allocator. | |
678 | endchoice | |
679 | ||
680 | config ZSWAP_ZPOOL_DEFAULT | |
681 | string | |
682 | depends on ZSWAP | |
683 | default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD | |
684 | default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD | |
685 | default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC | |
686 | default "" | |
687 | ||
688 | config ZSWAP_DEFAULT_ON | |
689 | bool "Enable the compressed cache for swap pages by default" | |
690 | depends on ZSWAP | |
691 | help | |
692 | If selected, the compressed cache for swap pages will be enabled | |
693 | at boot, otherwise it will be disabled. | |
694 | ||
695 | The selection made here can be overridden by using the kernel | |
696 | command line 'zswap.enabled=' option. | |
697 | ||
af8d417a DS |
698 | config ZPOOL |
699 | tristate "Common API for compressed memory storage" | |
0f8975ec | 700 | help |
af8d417a DS |
701 | Compressed memory storage API. This allows using either zbud or |
702 | zsmalloc. | |
0f8975ec | 703 | |
af8d417a | 704 | config ZBUD |
9a001fc1 | 705 | tristate "Low (Up to 2x) density storage for compressed pages" |
af8d417a DS |
706 | help |
707 | A special purpose allocator for storing compressed pages. | |
708 | It is designed to store up to two compressed pages per physical | |
709 | page. While this design limits storage density, it has simple and | |
710 | deterministic reclaim properties that make it preferable to a higher | |
711 | density approach when reclaim will be used. | |
bcf1647d | 712 | |
9a001fc1 VW |
713 | config Z3FOLD |
714 | tristate "Up to 3x density storage for compressed pages" | |
715 | depends on ZPOOL | |
9a001fc1 VW |
716 | help |
717 | A special purpose allocator for storing compressed pages. | |
718 | It is designed to store up to three compressed pages per physical | |
719 | page. It is a ZBUD derivative so the simplicity and determinism are | |
720 | still there. | |
721 | ||
bcf1647d | 722 | config ZSMALLOC |
d867f203 | 723 | tristate "Memory allocator for compressed pages" |
bcf1647d | 724 | depends on MMU |
bcf1647d MK |
725 | help |
726 | zsmalloc is a slab-based memory allocator designed to store | |
727 | compressed RAM pages. zsmalloc uses virtual memory mapping | |
728 | in order to reduce fragmentation. However, this results in a | |
729 | non-standard allocator interface where a handle, not a pointer, is | |
730 | returned by an alloc(). This handle must be mapped in order to | |
731 | access the allocated space. | |
732 | ||
0f050d99 GM |
733 | config ZSMALLOC_STAT |
734 | bool "Export zsmalloc statistics" | |
735 | depends on ZSMALLOC | |
736 | select DEBUG_FS | |
737 | help | |
738 | This option enables code in the zsmalloc to collect various | |
01ab1ede | 739 | statistics about what's happening in zsmalloc and exports that |
0f050d99 GM |
740 | information to userspace via debugfs. |
741 | If unsure, say N. | |
742 | ||
9e5c33d7 MS |
743 | config GENERIC_EARLY_IOREMAP |
744 | bool | |
042d27ac | 745 | |
22ee3ea5 HD |
746 | config STACK_MAX_DEFAULT_SIZE_MB |
747 | int "Default maximum user stack size for 32-bit processes (MB)" | |
748 | default 100 | |
042d27ac HD |
749 | range 8 2048 |
750 | depends on STACK_GROWSUP && (!64BIT || COMPAT) | |
751 | help | |
752 | This is the maximum stack size in Megabytes in the VM layout of 32-bit | |
753 | user processes when the stack grows upwards (currently only on parisc | |
22ee3ea5 | 754 | arch) when the RLIMIT_STACK hard limit is unlimited. |
042d27ac | 755 | |
22ee3ea5 | 756 | A sane initial value is 100 MB. |
3a80a7fa | 757 | |
3a80a7fa | 758 | config DEFERRED_STRUCT_PAGE_INIT |
1ce22103 | 759 | bool "Defer initialisation of struct pages to kthreads" |
d39f8fb4 | 760 | depends on SPARSEMEM |
ab1e8d89 | 761 | depends on !NEED_PER_CPU_KM |
889c695d | 762 | depends on 64BIT |
e4443149 | 763 | select PADATA |
3a80a7fa MG |
764 | help |
765 | Ordinarily all struct pages are initialised during early boot in a | |
766 | single thread. On very large machines this can take a considerable | |
767 | amount of time. If this option is set, large machines will bring up | |
e4443149 DJ |
768 | a subset of memmap at boot and then initialise the rest in parallel. |
769 | This has a potential performance impact on tasks running early in the | |
1ce22103 VB |
770 | lifetime of the system until these kthreads finish the |
771 | initialisation. | |
033fbae9 | 772 | |
33c3fc71 VD |
773 | config IDLE_PAGE_TRACKING |
774 | bool "Enable idle page tracking" | |
775 | depends on SYSFS && MMU | |
776 | select PAGE_EXTENSION if !64BIT | |
777 | help | |
778 | This feature allows to estimate the amount of user pages that have | |
779 | not been touched during a given period of time. This information can | |
780 | be useful to tune memory cgroup limits and/or for job placement | |
781 | within a compute cluster. | |
782 | ||
1ad1335d MR |
783 | See Documentation/admin-guide/mm/idle_page_tracking.rst for |
784 | more details. | |
33c3fc71 | 785 | |
c2280be8 AK |
786 | config ARCH_HAS_CACHE_LINE_SIZE |
787 | bool | |
788 | ||
17596731 | 789 | config ARCH_HAS_PTE_DEVMAP |
65f7d049 OH |
790 | bool |
791 | ||
033fbae9 | 792 | config ZONE_DEVICE |
5042db43 | 793 | bool "Device memory (pmem, HMM, etc...) hotplug support" |
033fbae9 DW |
794 | depends on MEMORY_HOTPLUG |
795 | depends on MEMORY_HOTREMOVE | |
99490f16 | 796 | depends on SPARSEMEM_VMEMMAP |
17596731 | 797 | depends on ARCH_HAS_PTE_DEVMAP |
3a08cd52 | 798 | select XARRAY_MULTI |
033fbae9 DW |
799 | |
800 | help | |
801 | Device memory hotplug support allows for establishing pmem, | |
802 | or other device driver discovered memory regions, in the | |
803 | memmap. This allows pfn_to_page() lookups of otherwise | |
804 | "device-physical" addresses which is needed for using a DAX | |
805 | mapping in an O_DIRECT operation, among other things. | |
806 | ||
807 | If FS_DAX is enabled, then say Y. | |
06a660ad | 808 | |
e7638488 DW |
809 | config DEV_PAGEMAP_OPS |
810 | bool | |
811 | ||
9c240a7b CH |
812 | # |
813 | # Helpers to mirror range of the CPU page tables of a process into device page | |
814 | # tables. | |
815 | # | |
c0b12405 | 816 | config HMM_MIRROR |
9c240a7b | 817 | bool |
f442c283 | 818 | depends on MMU |
c0b12405 | 819 | |
5042db43 JG |
820 | config DEVICE_PRIVATE |
821 | bool "Unaddressable device memory (GPU memory, ...)" | |
7328d9cc | 822 | depends on ZONE_DEVICE |
e7638488 | 823 | select DEV_PAGEMAP_OPS |
5042db43 JG |
824 | |
825 | help | |
826 | Allows creation of struct pages to represent unaddressable device | |
827 | memory; i.e., memory that is only accessible from the device (or | |
828 | group of devices). You likely also want to select HMM_MIRROR. | |
829 | ||
3e9a9e25 CH |
830 | config VMAP_PFN |
831 | bool | |
832 | ||
63c17fb8 DH |
833 | config ARCH_USES_HIGH_VMA_FLAGS |
834 | bool | |
66d37570 DH |
835 | config ARCH_HAS_PKEYS |
836 | bool | |
30a5b536 DZ |
837 | |
838 | config PERCPU_STATS | |
839 | bool "Collect percpu memory statistics" | |
30a5b536 DZ |
840 | help |
841 | This feature collects and exposes statistics via debugfs. The | |
842 | information includes global and per chunk statistics, which can | |
843 | be used to help understand percpu memory usage. | |
64c349f4 | 844 | |
9c84f229 JH |
845 | config GUP_TEST |
846 | bool "Enable infrastructure for get_user_pages()-related unit tests" | |
d0de8241 | 847 | depends on DEBUG_FS |
64c349f4 | 848 | help |
9c84f229 JH |
849 | Provides /sys/kernel/debug/gup_test, which in turn provides a way |
850 | to make ioctl calls that can launch kernel-based unit tests for | |
851 | the get_user_pages*() and pin_user_pages*() family of API calls. | |
64c349f4 | 852 | |
9c84f229 JH |
853 | These tests include benchmark testing of the _fast variants of |
854 | get_user_pages*() and pin_user_pages*(), as well as smoke tests of | |
855 | the non-_fast variants. | |
856 | ||
f4f9bda4 JH |
857 | There is also a sub-test that allows running dump_page() on any |
858 | of up to eight pages (selected by command line args) within the | |
859 | range of user-space addresses. These pages are either pinned via | |
860 | pin_user_pages*(), or pinned via get_user_pages*(), as specified | |
861 | by other command line arguments. | |
862 | ||
9c84f229 | 863 | See tools/testing/selftests/vm/gup_test.c |
3010a5ea | 864 | |
d0de8241 BS |
865 | comment "GUP_TEST needs to have DEBUG_FS enabled" |
866 | depends on !GUP_TEST && !DEBUG_FS | |
3010a5ea | 867 | |
39656e83 CH |
868 | config GUP_GET_PTE_LOW_HIGH |
869 | bool | |
870 | ||
99cb0dbd SL |
871 | config READ_ONLY_THP_FOR_FS |
872 | bool "Read-only THP for filesystems (EXPERIMENTAL)" | |
396bcc52 | 873 | depends on TRANSPARENT_HUGEPAGE && SHMEM |
99cb0dbd SL |
874 | |
875 | help | |
876 | Allow khugepaged to put read-only file-backed pages in THP. | |
877 | ||
878 | This is marked experimental because it is a new feature. Write | |
879 | support of file THPs will be developed in the next few release | |
880 | cycles. | |
881 | ||
3010a5ea LD |
882 | config ARCH_HAS_PTE_SPECIAL |
883 | bool | |
59e0b520 | 884 | |
cbd34da7 CH |
885 | # |
886 | # Some architectures require a special hugepage directory format that is | |
887 | # required to support multiple hugepage sizes. For example a4fe3ce76 | |
888 | # "powerpc/mm: Allow more flexible layouts for hugepage pagetables" | |
889 | # introduced it on powerpc. This allows for a more flexible hugepage | |
890 | # pagetable layouts. | |
891 | # | |
892 | config ARCH_HAS_HUGEPD | |
893 | bool | |
894 | ||
c5acad84 TH |
895 | config MAPPING_DIRTY_HELPERS |
896 | bool | |
897 | ||
298fa1ad TG |
898 | config KMAP_LOCAL |
899 | bool | |
900 | ||
1fbaf8fc CH |
901 | # struct io_mapping based helper. Selected by drivers that need them |
902 | config IO_MAPPING | |
903 | bool | |
59e0b520 | 904 | endmenu |