1 # SPDX-License-Identifier: GPL-2.0-only 1 # SPDX-License-Identifier: GPL-2.0-only
2 2
3 menu "Memory Management options" 3 menu "Memory Management options"
4 4
5 # 5 #
6 # For some reason microblaze and nios2 hard co 6 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
7 # add proper SWAP support to them, in which ca 7 # add proper SWAP support to them, in which case this can be remove.
8 # 8 #
9 config ARCH_NO_SWAP 9 config ARCH_NO_SWAP
10 bool 10 bool
11 11
12 config ZPOOL 12 config ZPOOL
13 bool 13 bool
14 14
15 menuconfig SWAP 15 menuconfig SWAP
16 bool "Support for paging of anonymous 16 bool "Support for paging of anonymous memory (swap)"
17 depends on MMU && BLOCK && !ARCH_NO_SW 17 depends on MMU && BLOCK && !ARCH_NO_SWAP
18 default y 18 default y
19 help 19 help
20 This option allows you to choose whe 20 This option allows you to choose whether you want to have support
21 for so called swap devices or swap f 21 for so called swap devices or swap files in your kernel that are
22 used to provide more virtual memory 22 used to provide more virtual memory than the actual RAM present
23 in your computer. If unsure say Y. 23 in your computer. If unsure say Y.
24 24
25 config ZSWAP 25 config ZSWAP
26 bool "Compressed cache for swap pages" 26 bool "Compressed cache for swap pages"
27 depends on SWAP 27 depends on SWAP
>> 28 select FRONTSWAP
28 select CRYPTO 29 select CRYPTO
29 select ZPOOL 30 select ZPOOL
30 help 31 help
31 A lightweight compressed cache for s 32 A lightweight compressed cache for swap pages. It takes
32 pages that are in the process of bei 33 pages that are in the process of being swapped out and attempts to
33 compress them into a dynamically all 34 compress them into a dynamically allocated RAM-based memory pool.
34 This can result in a significant I/O 35 This can result in a significant I/O reduction on swap device and,
35 in the case where decompressing from 36 in the case where decompressing from RAM is faster than swap device
36 reads, can also improve workload per 37 reads, can also improve workload performance.
37 38
38 config ZSWAP_DEFAULT_ON 39 config ZSWAP_DEFAULT_ON
39 bool "Enable the compressed cache for 40 bool "Enable the compressed cache for swap pages by default"
40 depends on ZSWAP 41 depends on ZSWAP
41 help 42 help
42 If selected, the compressed cache fo 43 If selected, the compressed cache for swap pages will be enabled
43 at boot, otherwise it will be disabl 44 at boot, otherwise it will be disabled.
44 45
45 The selection made here can be overr 46 The selection made here can be overridden by using the kernel
46 command line 'zswap.enabled=' option 47 command line 'zswap.enabled=' option.
47 48
48 config ZSWAP_SHRINKER_DEFAULT_ON <<
49 bool "Shrink the zswap pool on memory <<
50 depends on ZSWAP <<
51 default n <<
52 help <<
53 If selected, the zswap shrinker will <<
54 stored in the zswap pool will become <<
55 written back to the backing swap dev <<
56 <<
57 This means that zswap writeback coul <<
58 not yet full, or the cgroup zswap li <<
59 reducing the chance that cold pages <<
60 and consume memory indefinitely. <<
61 <<
62 choice 49 choice
63 prompt "Default compressor" 50 prompt "Default compressor"
64 depends on ZSWAP 51 depends on ZSWAP
65 default ZSWAP_COMPRESSOR_DEFAULT_LZO 52 default ZSWAP_COMPRESSOR_DEFAULT_LZO
66 help 53 help
67 Selects the default compression algo 54 Selects the default compression algorithm for the compressed cache
68 for swap pages. 55 for swap pages.
69 56
70 For an overview what kind of perform 57 For an overview what kind of performance can be expected from
71 a particular compression algorithm p 58 a particular compression algorithm please refer to the benchmarks
72 available at the following LWN page: 59 available at the following LWN page:
73 https://lwn.net/Articles/751795/ 60 https://lwn.net/Articles/751795/
74 61
75 If in doubt, select 'LZO'. 62 If in doubt, select 'LZO'.
76 63
77 The selection made here can be overr 64 The selection made here can be overridden by using the kernel
78 command line 'zswap.compressor=' opt 65 command line 'zswap.compressor=' option.
79 66
80 config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE 67 config ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
81 bool "Deflate" 68 bool "Deflate"
82 select CRYPTO_DEFLATE 69 select CRYPTO_DEFLATE
83 help 70 help
84 Use the Deflate algorithm as the def 71 Use the Deflate algorithm as the default compression algorithm.
85 72
86 config ZSWAP_COMPRESSOR_DEFAULT_LZO 73 config ZSWAP_COMPRESSOR_DEFAULT_LZO
87 bool "LZO" 74 bool "LZO"
88 select CRYPTO_LZO 75 select CRYPTO_LZO
89 help 76 help
90 Use the LZO algorithm as the default 77 Use the LZO algorithm as the default compression algorithm.
91 78
92 config ZSWAP_COMPRESSOR_DEFAULT_842 79 config ZSWAP_COMPRESSOR_DEFAULT_842
93 bool "842" 80 bool "842"
94 select CRYPTO_842 81 select CRYPTO_842
95 help 82 help
96 Use the 842 algorithm as the default 83 Use the 842 algorithm as the default compression algorithm.
97 84
98 config ZSWAP_COMPRESSOR_DEFAULT_LZ4 85 config ZSWAP_COMPRESSOR_DEFAULT_LZ4
99 bool "LZ4" 86 bool "LZ4"
100 select CRYPTO_LZ4 87 select CRYPTO_LZ4
101 help 88 help
102 Use the LZ4 algorithm as the default 89 Use the LZ4 algorithm as the default compression algorithm.
103 90
104 config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC 91 config ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
105 bool "LZ4HC" 92 bool "LZ4HC"
106 select CRYPTO_LZ4HC 93 select CRYPTO_LZ4HC
107 help 94 help
108 Use the LZ4HC algorithm as the defau 95 Use the LZ4HC algorithm as the default compression algorithm.
109 96
110 config ZSWAP_COMPRESSOR_DEFAULT_ZSTD 97 config ZSWAP_COMPRESSOR_DEFAULT_ZSTD
111 bool "zstd" 98 bool "zstd"
112 select CRYPTO_ZSTD 99 select CRYPTO_ZSTD
113 help 100 help
114 Use the zstd algorithm as the defaul 101 Use the zstd algorithm as the default compression algorithm.
115 endchoice 102 endchoice
116 103
117 config ZSWAP_COMPRESSOR_DEFAULT 104 config ZSWAP_COMPRESSOR_DEFAULT
118 string 105 string
119 depends on ZSWAP 106 depends on ZSWAP
120 default "deflate" if ZSWAP_COMPRESSOR_D 107 default "deflate" if ZSWAP_COMPRESSOR_DEFAULT_DEFLATE
121 default "lzo" if ZSWAP_COMPRESSOR_DEFAU 108 default "lzo" if ZSWAP_COMPRESSOR_DEFAULT_LZO
122 default "842" if ZSWAP_COMPRESSOR_DEFAU 109 default "842" if ZSWAP_COMPRESSOR_DEFAULT_842
123 default "lz4" if ZSWAP_COMPRESSOR_DEFAU 110 default "lz4" if ZSWAP_COMPRESSOR_DEFAULT_LZ4
124 default "lz4hc" if ZSWAP_COMPRESSOR_DEF 111 default "lz4hc" if ZSWAP_COMPRESSOR_DEFAULT_LZ4HC
125 default "zstd" if ZSWAP_COMPRESSOR_DEFA 112 default "zstd" if ZSWAP_COMPRESSOR_DEFAULT_ZSTD
126 default "" 113 default ""
127 114
128 choice 115 choice
129 prompt "Default allocator" 116 prompt "Default allocator"
130 depends on ZSWAP 117 depends on ZSWAP
131 default ZSWAP_ZPOOL_DEFAULT_ZSMALLOC i <<
132 default ZSWAP_ZPOOL_DEFAULT_ZBUD 118 default ZSWAP_ZPOOL_DEFAULT_ZBUD
133 help 119 help
134 Selects the default allocator for th 120 Selects the default allocator for the compressed cache for
135 swap pages. 121 swap pages.
136 The default is 'zbud' for compatibil 122 The default is 'zbud' for compatibility, however please do
137 read the description of each of the 123 read the description of each of the allocators below before
138 making a right choice. 124 making a right choice.
139 125
140 The selection made here can be overr 126 The selection made here can be overridden by using the kernel
141 command line 'zswap.zpool=' option. 127 command line 'zswap.zpool=' option.
142 128
143 config ZSWAP_ZPOOL_DEFAULT_ZBUD 129 config ZSWAP_ZPOOL_DEFAULT_ZBUD
144 bool "zbud" 130 bool "zbud"
145 select ZBUD 131 select ZBUD
146 help 132 help
147 Use the zbud allocator as the defaul 133 Use the zbud allocator as the default allocator.
148 134
149 config ZSWAP_ZPOOL_DEFAULT_Z3FOLD_DEPRECATED !! 135 config ZSWAP_ZPOOL_DEFAULT_Z3FOLD
150 bool "z3foldi (DEPRECATED)" !! 136 bool "z3fold"
151 select Z3FOLD_DEPRECATED !! 137 select Z3FOLD
152 help 138 help
153 Use the z3fold allocator as the defa 139 Use the z3fold allocator as the default allocator.
154 140
155 Deprecated and scheduled for removal <<
156 see CONFIG_Z3FOLD_DEPRECATED. <<
157 <<
158 config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC 141 config ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
159 bool "zsmalloc" 142 bool "zsmalloc"
160 select ZSMALLOC 143 select ZSMALLOC
161 help 144 help
162 Use the zsmalloc allocator as the de 145 Use the zsmalloc allocator as the default allocator.
163 endchoice 146 endchoice
164 147
165 config ZSWAP_ZPOOL_DEFAULT 148 config ZSWAP_ZPOOL_DEFAULT
166 string 149 string
167 depends on ZSWAP 150 depends on ZSWAP
168 default "zbud" if ZSWAP_ZPOOL_DEFAULT_Z 151 default "zbud" if ZSWAP_ZPOOL_DEFAULT_ZBUD
169 default "z3fold" if ZSWAP_ZPOOL_DEFAULT !! 152 default "z3fold" if ZSWAP_ZPOOL_DEFAULT_Z3FOLD
170 default "zsmalloc" if ZSWAP_ZPOOL_DEFAU 153 default "zsmalloc" if ZSWAP_ZPOOL_DEFAULT_ZSMALLOC
171 default "" 154 default ""
172 155
173 config ZBUD 156 config ZBUD
174 tristate "2:1 compression allocator (z 157 tristate "2:1 compression allocator (zbud)"
175 depends on ZSWAP 158 depends on ZSWAP
176 help 159 help
177 A special purpose allocator for stor 160 A special purpose allocator for storing compressed pages.
178 It is designed to store up to two co 161 It is designed to store up to two compressed pages per physical
179 page. While this design limits stor 162 page. While this design limits storage density, it has simple and
180 deterministic reclaim properties tha 163 deterministic reclaim properties that make it preferable to a higher
181 density approach when reclaim will b 164 density approach when reclaim will be used.
182 165
183 config Z3FOLD_DEPRECATED !! 166 config Z3FOLD
184 tristate "3:1 compression allocator (z !! 167 tristate "3:1 compression allocator (z3fold)"
185 depends on ZSWAP 168 depends on ZSWAP
186 help 169 help
187 Deprecated and scheduled for removal <<
188 a good reason for using Z3FOLD over <<
189 linux-mm@kvack.org and the zswap mai <<
190 <<
191 A special purpose allocator for stor 170 A special purpose allocator for storing compressed pages.
192 It is designed to store up to three 171 It is designed to store up to three compressed pages per physical
193 page. It is a ZBUD derivative so the 172 page. It is a ZBUD derivative so the simplicity and determinism are
194 still there. 173 still there.
195 174
196 config Z3FOLD <<
197 tristate <<
198 default y if Z3FOLD_DEPRECATED=y <<
199 default m if Z3FOLD_DEPRECATED=m <<
200 depends on Z3FOLD_DEPRECATED <<
201 <<
202 config ZSMALLOC 175 config ZSMALLOC
203 tristate 176 tristate
204 prompt "N:1 compression allocator (zsm !! 177 prompt "N:1 compression allocator (zsmalloc)" if ZSWAP
205 depends on MMU 178 depends on MMU
206 help 179 help
207 zsmalloc is a slab-based memory allo 180 zsmalloc is a slab-based memory allocator designed to store
208 pages of various compression levels 181 pages of various compression levels efficiently. It achieves
209 the highest storage density with the 182 the highest storage density with the least amount of fragmentation.
210 183
211 config ZSMALLOC_STAT 184 config ZSMALLOC_STAT
212 bool "Export zsmalloc statistics" 185 bool "Export zsmalloc statistics"
213 depends on ZSMALLOC 186 depends on ZSMALLOC
214 select DEBUG_FS 187 select DEBUG_FS
215 help 188 help
216 This option enables code in the zsma 189 This option enables code in the zsmalloc to collect various
217 statistics about what's happening in 190 statistics about what's happening in zsmalloc and exports that
218 information to userspace via debugfs 191 information to userspace via debugfs.
219 If unsure, say N. 192 If unsure, say N.
220 193
221 config ZSMALLOC_CHAIN_SIZE 194 config ZSMALLOC_CHAIN_SIZE
222 int "Maximum number of physical pages 195 int "Maximum number of physical pages per-zspage"
223 default 8 196 default 8
224 range 4 16 197 range 4 16
225 depends on ZSMALLOC 198 depends on ZSMALLOC
226 help 199 help
227 This option sets the upper limit on 200 This option sets the upper limit on the number of physical pages
228 that a zmalloc page (zspage) can con 201 that a zmalloc page (zspage) can consist of. The optimal zspage
229 chain size is calculated for each si 202 chain size is calculated for each size class during the
230 initialization of the pool. 203 initialization of the pool.
231 204
232 Changing this option can alter the c 205 Changing this option can alter the characteristics of size classes,
233 such as the number of pages per zspa 206 such as the number of pages per zspage and the number of objects
234 per zspage. This can also result in 207 per zspage. This can also result in different configurations of
235 the pool, as zsmalloc merges size cl 208 the pool, as zsmalloc merges size classes with similar
236 characteristics. 209 characteristics.
237 210
238 For more information, see zsmalloc d 211 For more information, see zsmalloc documentation.
239 212
240 menu "Slab allocator options" !! 213 menu "SLAB allocator options"
>> 214
>> 215 choice
>> 216 prompt "Choose SLAB allocator"
>> 217 default SLUB
>> 218 help
>> 219 This option allows to select a slab allocator.
>> 220
>> 221 config SLAB
>> 222 bool "SLAB"
>> 223 depends on !PREEMPT_RT
>> 224 select HAVE_HARDENED_USERCOPY_ALLOCATOR
>> 225 help
>> 226 The regular slab allocator that is established and known to work
>> 227 well in all environments. It organizes cache hot objects in
>> 228 per cpu and per node queues.
241 229
242 config SLUB 230 config SLUB
243 def_bool y !! 231 bool "SLUB (Unqueued Allocator)"
>> 232 select HAVE_HARDENED_USERCOPY_ALLOCATOR
>> 233 help
>> 234 SLUB is a slab allocator that minimizes cache line usage
>> 235 instead of managing queues of cached objects (SLAB approach).
>> 236 Per cpu caching is realized using slabs of objects instead
>> 237 of queues of objects. SLUB can use memory efficiently
>> 238 and has enhanced diagnostics. SLUB is the default choice for
>> 239 a slab allocator.
>> 240
>> 241 endchoice
244 242
245 config SLUB_TINY 243 config SLUB_TINY
246 bool "Configure for minimal memory foo !! 244 bool "Configure SLUB for minimal memory footprint"
247 depends on EXPERT !! 245 depends on SLUB && EXPERT
248 select SLAB_MERGE_DEFAULT 246 select SLAB_MERGE_DEFAULT
249 help 247 help
250 Configures the slab allocator in a !! 248 Configures the SLUB allocator in a way to achieve minimal memory
251 footprint, sacrificing scalability, 249 footprint, sacrificing scalability, debugging and other features.
252 This is intended only for the small 250 This is intended only for the smallest system that had used the
253 SLOB allocator and is not recommend 251 SLOB allocator and is not recommended for systems with more than
254 16MB RAM. 252 16MB RAM.
255 253
256 If unsure, say N. 254 If unsure, say N.
257 255
258 config SLAB_MERGE_DEFAULT 256 config SLAB_MERGE_DEFAULT
259 bool "Allow slab caches to be merged" 257 bool "Allow slab caches to be merged"
260 default y 258 default y
>> 259 depends on SLAB || SLUB
261 help 260 help
262 For reduced kernel memory fragmentat 261 For reduced kernel memory fragmentation, slab caches can be
263 merged when they share the same size 262 merged when they share the same size and other characteristics.
264 This carries a risk of kernel heap o 263 This carries a risk of kernel heap overflows being able to
265 overwrite objects from merged caches 264 overwrite objects from merged caches (and more easily control
266 cache layout), which makes such heap 265 cache layout), which makes such heap attacks easier to exploit
267 by attackers. By keeping caches unme 266 by attackers. By keeping caches unmerged, these kinds of exploits
268 can usually only damage objects in t 267 can usually only damage objects in the same cache. To disable
269 merging at runtime, "slab_nomerge" c 268 merging at runtime, "slab_nomerge" can be passed on the kernel
270 command line. 269 command line.
271 270
272 config SLAB_FREELIST_RANDOM 271 config SLAB_FREELIST_RANDOM
273 bool "Randomize slab freelist" 272 bool "Randomize slab freelist"
274 depends on !SLUB_TINY !! 273 depends on SLAB || (SLUB && !SLUB_TINY)
275 help 274 help
276 Randomizes the freelist order used o 275 Randomizes the freelist order used on creating new pages. This
277 security feature reduces the predict 276 security feature reduces the predictability of the kernel slab
278 allocator against heap overflows. 277 allocator against heap overflows.
279 278
280 config SLAB_FREELIST_HARDENED 279 config SLAB_FREELIST_HARDENED
281 bool "Harden slab freelist metadata" 280 bool "Harden slab freelist metadata"
282 depends on !SLUB_TINY !! 281 depends on SLAB || (SLUB && !SLUB_TINY)
283 help 282 help
284 Many kernel heap attacks try to targ 283 Many kernel heap attacks try to target slab cache metadata and
285 other infrastructure. This options m 284 other infrastructure. This options makes minor performance
286 sacrifices to harden the kernel slab 285 sacrifices to harden the kernel slab allocator against common
287 freelist exploit methods. !! 286 freelist exploit methods. Some slab implementations have more
288 !! 287 sanity-checking than others. This option is most effective with
289 config SLAB_BUCKETS !! 288 CONFIG_SLUB.
290 bool "Support allocation from separate <<
291 depends on !SLUB_TINY <<
292 default SLAB_FREELIST_HARDENED <<
293 help <<
294 Kernel heap attacks frequently depen <<
295 specifically-sized allocations with <<
296 that will be allocated into the same <<
297 target object. To avoid sharing thes <<
298 provide an explicitly separated set <<
299 user-controlled allocations. This ma <<
300 memory fragmentation, though in prac <<
301 of extra pages since the bulk of use <<
302 are relatively long-lived. <<
303 <<
304 If unsure, say Y. <<
305 289
306 config SLUB_STATS 290 config SLUB_STATS
307 default n 291 default n
308 bool "Enable performance statistics" !! 292 bool "Enable SLUB performance statistics"
309 depends on SYSFS && !SLUB_TINY !! 293 depends on SLUB && SYSFS && !SLUB_TINY
310 help 294 help
311 The statistics are useful to debug s !! 295 SLUB statistics are useful to debug SLUBs allocation behavior in
312 order find ways to optimize the allo 296 order find ways to optimize the allocator. This should never be
313 enabled for production use since kee 297 enabled for production use since keeping statistics slows down
314 the allocator by a few percentage po 298 the allocator by a few percentage points. The slabinfo command
315 supports the determination of the mo 299 supports the determination of the most active slabs to figure
316 out which slabs are relevant to a pa 300 out which slabs are relevant to a particular load.
317 Try running: slabinfo -DA 301 Try running: slabinfo -DA
318 302
319 config SLUB_CPU_PARTIAL 303 config SLUB_CPU_PARTIAL
320 default y 304 default y
321 depends on SMP && !SLUB_TINY !! 305 depends on SLUB && SMP && !SLUB_TINY
322 bool "Enable per cpu partial caches" !! 306 bool "SLUB per cpu partial cache"
323 help 307 help
324 Per cpu partial caches accelerate ob 308 Per cpu partial caches accelerate objects allocation and freeing
325 that is local to a processor at the 309 that is local to a processor at the price of more indeterminism
326 in the latency of the free. On overf 310 in the latency of the free. On overflow these caches will be cleared
327 which requires the taking of locks t 311 which requires the taking of locks that may cause latency spikes.
328 Typically one would choose no for a 312 Typically one would choose no for a realtime system.
329 313
330 config RANDOM_KMALLOC_CACHES !! 314 endmenu # SLAB allocator options
331 default n <<
332 depends on !SLUB_TINY <<
333 bool "Randomize slab caches for normal <<
334 help <<
335 A hardening feature that creates mul <<
336 normal kmalloc allocation and makes <<
337 on code address, which makes the att <<
338 vulnerable memory objects on the hea <<
339 memory vulnerabilities. <<
340 <<
341 Currently the number of copies is se <<
342 that effectively diverges the memory <<
343 subsystems or modules into different <<
344 limited degree of memory and CPU ove <<
345 system workload. <<
346 <<
347 endmenu # Slab allocator options <<
348 315
349 config SHUFFLE_PAGE_ALLOCATOR 316 config SHUFFLE_PAGE_ALLOCATOR
350 bool "Page allocator randomization" 317 bool "Page allocator randomization"
351 default SLAB_FREELIST_RANDOM && ACPI_N 318 default SLAB_FREELIST_RANDOM && ACPI_NUMA
352 help 319 help
353 Randomization of the page allocator 320 Randomization of the page allocator improves the average
354 utilization of a direct-mapped memor 321 utilization of a direct-mapped memory-side-cache. See section
355 5.2.27 Heterogeneous Memory Attribut 322 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
356 6.2a specification for an example of 323 6.2a specification for an example of how a platform advertises
357 the presence of a memory-side-cache. 324 the presence of a memory-side-cache. There are also incidental
358 security benefits as it reduces the 325 security benefits as it reduces the predictability of page
359 allocations to compliment SLAB_FREEL 326 allocations to compliment SLAB_FREELIST_RANDOM, but the
360 default granularity of shuffling on !! 327 default granularity of shuffling on the MAX_ORDER i.e, 10th
361 order of pages is selected based on 328 order of pages is selected based on cache utilization benefits
362 on x86. 329 on x86.
363 330
364 While the randomization improves cac 331 While the randomization improves cache utilization it may
365 negatively impact workloads on platf 332 negatively impact workloads on platforms without a cache. For
366 this reason, by default, the randomi !! 333 this reason, by default, the randomization is enabled only
367 if SHUFFLE_PAGE_ALLOCATOR=y. The ran !! 334 after runtime detection of a direct-mapped memory-side-cache.
368 with the 'page_alloc.shuffle' kernel !! 335 Otherwise, the randomization may be force enabled with the
>> 336 'page_alloc.shuffle' kernel command line parameter.
369 337
370 Say Y if unsure. 338 Say Y if unsure.
371 339
372 config COMPAT_BRK 340 config COMPAT_BRK
373 bool "Disable heap randomization" 341 bool "Disable heap randomization"
374 default y 342 default y
375 help 343 help
376 Randomizing heap placement makes hea 344 Randomizing heap placement makes heap exploits harder, but it
377 also breaks ancient binaries (includ 345 also breaks ancient binaries (including anything libc5 based).
378 This option changes the bootup defau 346 This option changes the bootup default to heap randomization
379 disabled, and can be overridden at r 347 disabled, and can be overridden at runtime by setting
380 /proc/sys/kernel/randomize_va_space 348 /proc/sys/kernel/randomize_va_space to 2.
381 349
382 On non-ancient distros (post-2000 on 350 On non-ancient distros (post-2000 ones) N is usually a safe choice.
383 351
384 config MMAP_ALLOW_UNINITIALIZED 352 config MMAP_ALLOW_UNINITIALIZED
385 bool "Allow mmapped anonymous memory t 353 bool "Allow mmapped anonymous memory to be uninitialized"
386 depends on EXPERT && !MMU 354 depends on EXPERT && !MMU
387 default n 355 default n
388 help 356 help
389 Normally, and according to the Linux 357 Normally, and according to the Linux spec, anonymous memory obtained
390 from mmap() has its contents cleared 358 from mmap() has its contents cleared before it is passed to
391 userspace. Enabling this config opt 359 userspace. Enabling this config option allows you to request that
392 mmap() skip that if it is given an M 360 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
393 providing a huge performance boost. 361 providing a huge performance boost. If this option is not enabled,
394 then the flag will be ignored. 362 then the flag will be ignored.
395 363
396 This is taken advantage of by uClibc 364 This is taken advantage of by uClibc's malloc(), and also by
397 ELF-FDPIC binfmt's brk and stack all 365 ELF-FDPIC binfmt's brk and stack allocator.
398 366
399 Because of the obvious security issu 367 Because of the obvious security issues, this option should only be
400 enabled on embedded devices where yo 368 enabled on embedded devices where you control what is run in
401 userspace. Since that isn't general 369 userspace. Since that isn't generally a problem on no-MMU systems,
402 it is normally safe to say Y here. 370 it is normally safe to say Y here.
403 371
404 See Documentation/admin-guide/mm/nom 372 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
405 373
406 config SELECT_MEMORY_MODEL 374 config SELECT_MEMORY_MODEL
407 def_bool y 375 def_bool y
408 depends on ARCH_SELECT_MEMORY_MODEL 376 depends on ARCH_SELECT_MEMORY_MODEL
409 377
410 choice 378 choice
411 prompt "Memory model" 379 prompt "Memory model"
412 depends on SELECT_MEMORY_MODEL 380 depends on SELECT_MEMORY_MODEL
413 default SPARSEMEM_MANUAL if ARCH_SPARS 381 default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT
414 default FLATMEM_MANUAL 382 default FLATMEM_MANUAL
415 help 383 help
416 This option allows you to change som 384 This option allows you to change some of the ways that
417 Linux manages its memory internally. 385 Linux manages its memory internally. Most users will
418 only have one option here selected b 386 only have one option here selected by the architecture
419 configuration. This is normal. 387 configuration. This is normal.
420 388
421 config FLATMEM_MANUAL 389 config FLATMEM_MANUAL
422 bool "Flat Memory" 390 bool "Flat Memory"
423 depends on !ARCH_SPARSEMEM_ENABLE || A 391 depends on !ARCH_SPARSEMEM_ENABLE || ARCH_FLATMEM_ENABLE
424 help 392 help
425 This option is best suited for non-N 393 This option is best suited for non-NUMA systems with
426 flat address space. The FLATMEM is t 394 flat address space. The FLATMEM is the most efficient
427 system in terms of performance and r 395 system in terms of performance and resource consumption
428 and it is the best option for smalle 396 and it is the best option for smaller systems.
429 397
430 For systems that have holes in their 398 For systems that have holes in their physical address
431 spaces and for features like NUMA an 399 spaces and for features like NUMA and memory hotplug,
432 choose "Sparse Memory". 400 choose "Sparse Memory".
433 401
434 If unsure, choose this option (Flat 402 If unsure, choose this option (Flat Memory) over any other.
435 403
436 config SPARSEMEM_MANUAL 404 config SPARSEMEM_MANUAL
437 bool "Sparse Memory" 405 bool "Sparse Memory"
438 depends on ARCH_SPARSEMEM_ENABLE 406 depends on ARCH_SPARSEMEM_ENABLE
439 help 407 help
440 This will be the only option for som 408 This will be the only option for some systems, including
441 memory hot-plug systems. This is no 409 memory hot-plug systems. This is normal.
442 410
443 This option provides efficient suppo 411 This option provides efficient support for systems with
444 holes is their physical address spac 412 holes is their physical address space and allows memory
445 hot-plug and hot-remove. 413 hot-plug and hot-remove.
446 414
447 If unsure, choose "Flat Memory" over 415 If unsure, choose "Flat Memory" over this option.
448 416
449 endchoice 417 endchoice
450 418
451 config SPARSEMEM 419 config SPARSEMEM
452 def_bool y 420 def_bool y
453 depends on (!SELECT_MEMORY_MODEL && AR 421 depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL
454 422
455 config FLATMEM 423 config FLATMEM
456 def_bool y 424 def_bool y
457 depends on !SPARSEMEM || FLATMEM_MANUA 425 depends on !SPARSEMEM || FLATMEM_MANUAL
458 426
459 # 427 #
460 # SPARSEMEM_EXTREME (which is the default) doe 428 # SPARSEMEM_EXTREME (which is the default) does some bootmem
461 # allocations when sparse_init() is called. I 429 # allocations when sparse_init() is called. If this cannot
462 # be done on your architecture, select this op 430 # be done on your architecture, select this option. However,
463 # statically allocating the mem_section[] arra 431 # statically allocating the mem_section[] array can potentially
464 # consume vast quantities of .bss, so be caref 432 # consume vast quantities of .bss, so be careful.
465 # 433 #
466 # This option will also potentially produce sm 434 # This option will also potentially produce smaller runtime code
467 # with gcc 3.4 and later. 435 # with gcc 3.4 and later.
468 # 436 #
469 config SPARSEMEM_STATIC 437 config SPARSEMEM_STATIC
470 bool 438 bool
471 439
472 # 440 #
473 # Architecture platforms which require a two l 441 # Architecture platforms which require a two level mem_section in SPARSEMEM
474 # must select this option. This is usually for 442 # must select this option. This is usually for architecture platforms with
475 # an extremely sparse physical address space. 443 # an extremely sparse physical address space.
476 # 444 #
477 config SPARSEMEM_EXTREME 445 config SPARSEMEM_EXTREME
478 def_bool y 446 def_bool y
479 depends on SPARSEMEM && !SPARSEMEM_STA 447 depends on SPARSEMEM && !SPARSEMEM_STATIC
480 448
481 config SPARSEMEM_VMEMMAP_ENABLE 449 config SPARSEMEM_VMEMMAP_ENABLE
482 bool 450 bool
483 451
484 config SPARSEMEM_VMEMMAP 452 config SPARSEMEM_VMEMMAP
485 bool "Sparse Memory virtual memmap" 453 bool "Sparse Memory virtual memmap"
486 depends on SPARSEMEM && SPARSEMEM_VMEM 454 depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE
487 default y 455 default y
488 help 456 help
489 SPARSEMEM_VMEMMAP uses a virtually m 457 SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise
490 pfn_to_page and page_to_pfn operatio 458 pfn_to_page and page_to_pfn operations. This is the most
491 efficient option when sufficient ker 459 efficient option when sufficient kernel resources are available.
492 # 460 #
493 # Select this config option from the architect 461 # Select this config option from the architecture Kconfig, if it is preferred
494 # to enable the feature of HugeTLB/dev_dax vme 462 # to enable the feature of HugeTLB/dev_dax vmemmap optimization.
495 # 463 #
496 config ARCH_WANT_OPTIMIZE_DAX_VMEMMAP !! 464 config ARCH_WANT_OPTIMIZE_VMEMMAP
497 bool <<
498 <<
499 config ARCH_WANT_OPTIMIZE_HUGETLB_VMEMMAP <<
500 bool 465 bool
501 466
502 config HAVE_MEMBLOCK_PHYS_MAP 467 config HAVE_MEMBLOCK_PHYS_MAP
503 bool 468 bool
504 469
505 config HAVE_GUP_FAST !! 470 config HAVE_FAST_GUP
506 depends on MMU 471 depends on MMU
507 bool 472 bool
508 473
509 # Don't discard allocated memory used to track 474 # Don't discard allocated memory used to track "memory" and "reserved" memblocks
510 # after early boot, so it can still be used to 475 # after early boot, so it can still be used to test for validity of memory.
511 # Also, memblocks are updated with memory hot( 476 # Also, memblocks are updated with memory hot(un)plug.
512 config ARCH_KEEP_MEMBLOCK 477 config ARCH_KEEP_MEMBLOCK
513 bool 478 bool
514 479
515 # Keep arch NUMA mapping infrastructure post-i 480 # Keep arch NUMA mapping infrastructure post-init.
516 config NUMA_KEEP_MEMINFO 481 config NUMA_KEEP_MEMINFO
517 bool 482 bool
518 483
519 config MEMORY_ISOLATION 484 config MEMORY_ISOLATION
520 bool 485 bool
521 486
522 # IORESOURCE_SYSTEM_RAM regions in the kernel 487 # IORESOURCE_SYSTEM_RAM regions in the kernel resource tree that are marked
523 # IORESOURCE_EXCLUSIVE cannot be mapped to use 488 # IORESOURCE_EXCLUSIVE cannot be mapped to user space, for example, via
524 # /dev/mem. 489 # /dev/mem.
525 config EXCLUSIVE_SYSTEM_RAM 490 config EXCLUSIVE_SYSTEM_RAM
526 def_bool y 491 def_bool y
527 depends on !DEVMEM || STRICT_DEVMEM 492 depends on !DEVMEM || STRICT_DEVMEM
528 493
529 # 494 #
530 # Only be set on architectures that have compl 495 # Only be set on architectures that have completely implemented memory hotplug
531 # feature. If you are not sure, don't touch it 496 # feature. If you are not sure, don't touch it.
532 # 497 #
533 config HAVE_BOOTMEM_INFO_NODE 498 config HAVE_BOOTMEM_INFO_NODE
534 def_bool n 499 def_bool n
535 500
536 config ARCH_ENABLE_MEMORY_HOTPLUG 501 config ARCH_ENABLE_MEMORY_HOTPLUG
537 bool 502 bool
538 503
539 config ARCH_ENABLE_MEMORY_HOTREMOVE 504 config ARCH_ENABLE_MEMORY_HOTREMOVE
540 bool 505 bool
541 506
542 # eventually, we can have this option just 'se 507 # eventually, we can have this option just 'select SPARSEMEM'
543 menuconfig MEMORY_HOTPLUG 508 menuconfig MEMORY_HOTPLUG
544 bool "Memory hotplug" 509 bool "Memory hotplug"
545 select MEMORY_ISOLATION 510 select MEMORY_ISOLATION
546 depends on SPARSEMEM 511 depends on SPARSEMEM
547 depends on ARCH_ENABLE_MEMORY_HOTPLUG 512 depends on ARCH_ENABLE_MEMORY_HOTPLUG
548 depends on 64BIT 513 depends on 64BIT
549 select NUMA_KEEP_MEMINFO if NUMA 514 select NUMA_KEEP_MEMINFO if NUMA
550 515
551 if MEMORY_HOTPLUG 516 if MEMORY_HOTPLUG
552 517
553 config MEMORY_HOTPLUG_DEFAULT_ONLINE 518 config MEMORY_HOTPLUG_DEFAULT_ONLINE
554 bool "Online the newly added memory bl 519 bool "Online the newly added memory blocks by default"
555 depends on MEMORY_HOTPLUG 520 depends on MEMORY_HOTPLUG
556 help 521 help
557 This option sets the default policy 522 This option sets the default policy setting for memory hotplug
558 onlining policy (/sys/devices/system 523 onlining policy (/sys/devices/system/memory/auto_online_blocks) which
559 determines what happens to newly add 524 determines what happens to newly added memory regions. Policy setting
560 can always be changed at runtime. 525 can always be changed at runtime.
561 See Documentation/admin-guide/mm/mem 526 See Documentation/admin-guide/mm/memory-hotplug.rst for more information.
562 527
563 Say Y here if you want all hot-plugg 528 Say Y here if you want all hot-plugged memory blocks to appear in
564 'online' state by default. 529 'online' state by default.
565 Say N here if you want the default p 530 Say N here if you want the default policy to keep all hot-plugged
566 memory blocks in 'offline' state. 531 memory blocks in 'offline' state.
567 532
568 config MEMORY_HOTREMOVE 533 config MEMORY_HOTREMOVE
569 bool "Allow for memory hot remove" 534 bool "Allow for memory hot remove"
570 select HAVE_BOOTMEM_INFO_NODE if (X86_ 535 select HAVE_BOOTMEM_INFO_NODE if (X86_64 || PPC64)
571 depends on MEMORY_HOTPLUG && ARCH_ENAB 536 depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE
572 depends on MIGRATION 537 depends on MIGRATION
573 538
574 config MHP_MEMMAP_ON_MEMORY 539 config MHP_MEMMAP_ON_MEMORY
575 def_bool y 540 def_bool y
576 depends on MEMORY_HOTPLUG && SPARSEMEM 541 depends on MEMORY_HOTPLUG && SPARSEMEM_VMEMMAP
577 depends on ARCH_MHP_MEMMAP_ON_MEMORY_E 542 depends on ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
578 543
579 endif # MEMORY_HOTPLUG 544 endif # MEMORY_HOTPLUG
580 545
581 config ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE <<
582 bool <<
583 <<
584 # Heavily threaded applications may benefit fr 546 # Heavily threaded applications may benefit from splitting the mm-wide
585 # page_table_lock, so that faults on different 547 # page_table_lock, so that faults on different parts of the user address
586 # space can be handled with less contention: s 548 # space can be handled with less contention: split it at this NR_CPUS.
587 # Default to 4 for wider testing, though 8 mig 549 # Default to 4 for wider testing, though 8 might be more appropriate.
588 # ARM's adjust_pte (unused if VIPT) depends on 550 # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
589 # PA-RISC 7xxx's spinlock_t would enlarge stru 551 # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
590 # SPARC32 allocates multiple pte tables within 552 # SPARC32 allocates multiple pte tables within a single page, and therefore
591 # a per-page lock leads to problems when multi 553 # a per-page lock leads to problems when multiple tables need to be locked
592 # at the same time (e.g. copy_page_range()). 554 # at the same time (e.g. copy_page_range()).
593 # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock 555 # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page.
594 # 556 #
595 config SPLIT_PTE_PTLOCKS !! 557 config SPLIT_PTLOCK_CPUS
596 def_bool y !! 558 int
597 depends on MMU !! 559 default "999999" if !MMU
598 depends on SMP !! 560 default "999999" if ARM && !CPU_CACHE_VIPT
599 depends on NR_CPUS >= 4 !! 561 default "999999" if PARISC && !PA20
600 depends on !ARM || CPU_CACHE_VIPT !! 562 default "999999" if SPARC32
601 depends on !PARISC || PA20 !! 563 default "4"
602 depends on !SPARC32 <<
603 564
604 config ARCH_ENABLE_SPLIT_PMD_PTLOCK 565 config ARCH_ENABLE_SPLIT_PMD_PTLOCK
605 bool 566 bool
606 567
607 config SPLIT_PMD_PTLOCKS <<
608 def_bool y <<
609 depends on SPLIT_PTE_PTLOCKS && ARCH_E <<
610 <<
611 # 568 #
612 # support for memory balloon 569 # support for memory balloon
613 config MEMORY_BALLOON 570 config MEMORY_BALLOON
614 bool 571 bool
615 572
616 # 573 #
617 # support for memory balloon compaction 574 # support for memory balloon compaction
618 config BALLOON_COMPACTION 575 config BALLOON_COMPACTION
619 bool "Allow for balloon memory compact 576 bool "Allow for balloon memory compaction/migration"
620 default y !! 577 def_bool y
621 depends on COMPACTION && MEMORY_BALLOO 578 depends on COMPACTION && MEMORY_BALLOON
622 help 579 help
623 Memory fragmentation introduced by b 580 Memory fragmentation introduced by ballooning might reduce
624 significantly the number of 2MB cont 581 significantly the number of 2MB contiguous memory blocks that can be
625 used within a guest, thus imposing p 582 used within a guest, thus imposing performance penalties associated
626 with the reduced number of transpare 583 with the reduced number of transparent huge pages that could be used
627 by the guest workload. Allowing the 584 by the guest workload. Allowing the compaction & migration for memory
628 pages enlisted as being part of memo 585 pages enlisted as being part of memory balloon devices avoids the
629 scenario aforementioned and helps im 586 scenario aforementioned and helps improving memory defragmentation.
630 587
631 # 588 #
632 # support for memory compaction 589 # support for memory compaction
633 config COMPACTION 590 config COMPACTION
634 bool "Allow for memory compaction" 591 bool "Allow for memory compaction"
635 default y !! 592 def_bool y
636 select MIGRATION 593 select MIGRATION
637 depends on MMU 594 depends on MMU
638 help 595 help
639 Compaction is the only memory manage 596 Compaction is the only memory management component to form
640 high order (larger physically contig 597 high order (larger physically contiguous) memory blocks
641 reliably. The page allocator relies 598 reliably. The page allocator relies on compaction heavily and
642 the lack of the feature can lead to 599 the lack of the feature can lead to unexpected OOM killer
643 invocations for high order memory re 600 invocations for high order memory requests. You shouldn't
644 disable this option unless there rea 601 disable this option unless there really is a strong reason for
645 it and then we would be really inter 602 it and then we would be really interested to hear about that at
646 linux-mm@kvack.org. 603 linux-mm@kvack.org.
647 604
648 config COMPACT_UNEVICTABLE_DEFAULT 605 config COMPACT_UNEVICTABLE_DEFAULT
649 int 606 int
650 depends on COMPACTION 607 depends on COMPACTION
651 default 0 if PREEMPT_RT 608 default 0 if PREEMPT_RT
652 default 1 609 default 1
653 610
654 # 611 #
655 # support for free page reporting 612 # support for free page reporting
656 config PAGE_REPORTING 613 config PAGE_REPORTING
657 bool "Free page reporting" 614 bool "Free page reporting"
>> 615 def_bool n
658 help 616 help
659 Free page reporting allows for the i 617 Free page reporting allows for the incremental acquisition of
660 free pages from the buddy allocator 618 free pages from the buddy allocator for the purpose of reporting
661 those pages to another entity, such 619 those pages to another entity, such as a hypervisor, so that the
662 memory can be freed within the host 620 memory can be freed within the host for other uses.
663 621
664 # 622 #
665 # support for page migration 623 # support for page migration
666 # 624 #
667 config MIGRATION 625 config MIGRATION
668 bool "Page migration" 626 bool "Page migration"
669 default y !! 627 def_bool y
670 depends on (NUMA || ARCH_ENABLE_MEMORY 628 depends on (NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION || CMA) && MMU
671 help 629 help
672 Allows the migration of the physical 630 Allows the migration of the physical location of pages of processes
673 while the virtual addresses are not 631 while the virtual addresses are not changed. This is useful in
674 two situations. The first is on NUMA 632 two situations. The first is on NUMA systems to put pages nearer
675 to the processors accessing. The sec 633 to the processors accessing. The second is when allocating huge
676 pages as migration can relocate page 634 pages as migration can relocate pages to satisfy a huge page
677 allocation instead of reclaiming. 635 allocation instead of reclaiming.
678 636
679 config DEVICE_MIGRATION 637 config DEVICE_MIGRATION
680 def_bool MIGRATION && ZONE_DEVICE 638 def_bool MIGRATION && ZONE_DEVICE
681 639
682 config ARCH_ENABLE_HUGEPAGE_MIGRATION 640 config ARCH_ENABLE_HUGEPAGE_MIGRATION
683 bool 641 bool
684 642
685 config ARCH_ENABLE_THP_MIGRATION 643 config ARCH_ENABLE_THP_MIGRATION
686 bool 644 bool
687 645
688 config HUGETLB_PAGE_SIZE_VARIABLE 646 config HUGETLB_PAGE_SIZE_VARIABLE
689 def_bool n 647 def_bool n
690 help 648 help
691 Allows the pageblock_order value to 649 Allows the pageblock_order value to be dynamic instead of just standard
692 HUGETLB_PAGE_ORDER when there are mu 650 HUGETLB_PAGE_ORDER when there are multiple HugeTLB page sizes available
693 on a platform. 651 on a platform.
694 652
695 Note that the pageblock_order cannot !! 653 Note that the pageblock_order cannot exceed MAX_ORDER and will be
696 clamped down to MAX_PAGE_ORDER. !! 654 clamped down to MAX_ORDER.
697 655
698 config CONTIG_ALLOC 656 config CONTIG_ALLOC
699 def_bool (MEMORY_ISOLATION && COMPACTI 657 def_bool (MEMORY_ISOLATION && COMPACTION) || CMA
700 658
701 config PCP_BATCH_SCALE_MAX <<
702 int "Maximum scale factor of PCP (Per- <<
703 default 5 <<
704 range 0 6 <<
705 help <<
706 In page allocator, PCP (Per-CPU page <<
707 batches. The batch number is scaled <<
708 allocation/free throughput. But too <<
709 latency. This option sets the upper <<
710 the maximum latency. <<
711 <<
712 config PHYS_ADDR_T_64BIT 659 config PHYS_ADDR_T_64BIT
713 def_bool 64BIT 660 def_bool 64BIT
714 661
715 config BOUNCE 662 config BOUNCE
716 bool "Enable bounce buffers" 663 bool "Enable bounce buffers"
717 default y 664 default y
718 depends on BLOCK && MMU && HIGHMEM 665 depends on BLOCK && MMU && HIGHMEM
719 help 666 help
720 Enable bounce buffers for devices th 667 Enable bounce buffers for devices that cannot access the full range of
721 memory available to the CPU. Enabled 668 memory available to the CPU. Enabled by default when HIGHMEM is
722 selected, but you may say n to overr 669 selected, but you may say n to override this.
723 670
724 config MMU_NOTIFIER 671 config MMU_NOTIFIER
725 bool 672 bool
726 select INTERVAL_TREE 673 select INTERVAL_TREE
727 674
728 config KSM 675 config KSM
729 bool "Enable KSM for page merging" 676 bool "Enable KSM for page merging"
730 depends on MMU 677 depends on MMU
731 select XXHASH 678 select XXHASH
732 help 679 help
733 Enable Kernel Samepage Merging: KSM 680 Enable Kernel Samepage Merging: KSM periodically scans those areas
734 of an application's address space th 681 of an application's address space that an app has advised may be
735 mergeable. When it finds pages of i 682 mergeable. When it finds pages of identical content, it replaces
736 the many instances by a single page 683 the many instances by a single page with that content, so
737 saving memory until one or another a 684 saving memory until one or another app needs to modify the content.
738 Recommended for use with KVM, or wit 685 Recommended for use with KVM, or with other duplicative applications.
739 See Documentation/mm/ksm.rst for mor 686 See Documentation/mm/ksm.rst for more information: KSM is inactive
740 until a program has madvised that an 687 until a program has madvised that an area is MADV_MERGEABLE, and
741 root has set /sys/kernel/mm/ksm/run 688 root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
742 689
743 config DEFAULT_MMAP_MIN_ADDR 690 config DEFAULT_MMAP_MIN_ADDR
744 int "Low address space to protect from 691 int "Low address space to protect from user allocation"
745 depends on MMU 692 depends on MMU
746 default 4096 693 default 4096
747 help 694 help
748 This is the portion of low virtual m 695 This is the portion of low virtual memory which should be protected
749 from userspace allocation. Keeping 696 from userspace allocation. Keeping a user from writing to low pages
750 can help reduce the impact of kernel 697 can help reduce the impact of kernel NULL pointer bugs.
751 698
752 For most arm64, ppc64 and x86 users !! 699 For most ia64, ppc64 and x86 users with lots of address space
753 a value of 65536 is reasonable and s 700 a value of 65536 is reasonable and should cause no problems.
754 On arm and other archs it should not 701 On arm and other archs it should not be higher than 32768.
755 Programs which use vm86 functionalit 702 Programs which use vm86 functionality or have some need to map
756 this low address space will need CAP 703 this low address space will need CAP_SYS_RAWIO or disable this
757 protection by setting the value to 0 704 protection by setting the value to 0.
758 705
759 This value can be changed after boot 706 This value can be changed after boot using the
760 /proc/sys/vm/mmap_min_addr tunable. 707 /proc/sys/vm/mmap_min_addr tunable.
761 708
762 config ARCH_SUPPORTS_MEMORY_FAILURE 709 config ARCH_SUPPORTS_MEMORY_FAILURE
763 bool 710 bool
764 711
765 config MEMORY_FAILURE 712 config MEMORY_FAILURE
766 depends on MMU 713 depends on MMU
767 depends on ARCH_SUPPORTS_MEMORY_FAILUR 714 depends on ARCH_SUPPORTS_MEMORY_FAILURE
768 bool "Enable recovery from hardware me 715 bool "Enable recovery from hardware memory errors"
769 select MEMORY_ISOLATION 716 select MEMORY_ISOLATION
770 select RAS 717 select RAS
771 help 718 help
772 Enables code to recover from some me 719 Enables code to recover from some memory failures on systems
773 with MCA recovery. This allows a sys 720 with MCA recovery. This allows a system to continue running
774 even when some of its memory has unc 721 even when some of its memory has uncorrected errors. This requires
775 special hardware support and typical 722 special hardware support and typically ECC memory.
776 723
777 config HWPOISON_INJECT 724 config HWPOISON_INJECT
778 tristate "HWPoison pages injector" 725 tristate "HWPoison pages injector"
779 depends on MEMORY_FAILURE && DEBUG_KER 726 depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS
780 select PROC_PAGE_MONITOR 727 select PROC_PAGE_MONITOR
781 728
782 config NOMMU_INITIAL_TRIM_EXCESS 729 config NOMMU_INITIAL_TRIM_EXCESS
783 int "Turn on mmap() excess space trimm 730 int "Turn on mmap() excess space trimming before booting"
784 depends on !MMU 731 depends on !MMU
785 default 1 732 default 1
786 help 733 help
787 The NOMMU mmap() frequently needs to 734 The NOMMU mmap() frequently needs to allocate large contiguous chunks
788 of memory on which to store mappings 735 of memory on which to store mappings, but it can only ask the system
789 allocator for chunks in 2^N*PAGE_SIZ 736 allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
790 more than it requires. To deal with 737 more than it requires. To deal with this, mmap() is able to trim off
791 the excess and return it to the allo 738 the excess and return it to the allocator.
792 739
793 If trimming is enabled, the excess i 740 If trimming is enabled, the excess is trimmed off and returned to the
794 system allocator, which can cause ex 741 system allocator, which can cause extra fragmentation, particularly
795 if there are a lot of transient proc 742 if there are a lot of transient processes.
796 743
797 If trimming is disabled, the excess 744 If trimming is disabled, the excess is kept, but not used, which for
798 long-term mappings means that the sp 745 long-term mappings means that the space is wasted.
799 746
800 Trimming can be dynamically controll 747 Trimming can be dynamically controlled through a sysctl option
801 (/proc/sys/vm/nr_trim_pages) which s 748 (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of
802 excess pages there must be before tr 749 excess pages there must be before trimming should occur, or zero if
803 no trimming is to occur. 750 no trimming is to occur.
804 751
805 This option specifies the initial va 752 This option specifies the initial value of this option. The default
806 of 1 says that all excess pages shou 753 of 1 says that all excess pages should be trimmed.
807 754
808 See Documentation/admin-guide/mm/nom 755 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
809 756
810 config ARCH_WANT_GENERAL_HUGETLB 757 config ARCH_WANT_GENERAL_HUGETLB
811 bool 758 bool
812 759
813 config ARCH_WANTS_THP_SWAP 760 config ARCH_WANTS_THP_SWAP
814 def_bool n 761 def_bool n
815 762
816 menuconfig TRANSPARENT_HUGEPAGE 763 menuconfig TRANSPARENT_HUGEPAGE
817 bool "Transparent Hugepage Support" 764 bool "Transparent Hugepage Support"
818 depends on HAVE_ARCH_TRANSPARENT_HUGEP 765 depends on HAVE_ARCH_TRANSPARENT_HUGEPAGE && !PREEMPT_RT
819 select COMPACTION 766 select COMPACTION
820 select XARRAY_MULTI 767 select XARRAY_MULTI
821 help 768 help
822 Transparent Hugepages allows the ker 769 Transparent Hugepages allows the kernel to use huge pages and
823 huge tlb transparently to the applic 770 huge tlb transparently to the applications whenever possible.
824 This feature can improve computing p 771 This feature can improve computing performance to certain
825 applications by speeding up page fau 772 applications by speeding up page faults during memory
826 allocation, by reducing the number o 773 allocation, by reducing the number of tlb misses and by speeding
827 up the pagetable walking. 774 up the pagetable walking.
828 775
829 If memory constrained on embedded, y 776 If memory constrained on embedded, you may want to say N.
830 777
831 if TRANSPARENT_HUGEPAGE 778 if TRANSPARENT_HUGEPAGE
832 779
833 choice 780 choice
834 prompt "Transparent Hugepage Support s 781 prompt "Transparent Hugepage Support sysfs defaults"
835 depends on TRANSPARENT_HUGEPAGE 782 depends on TRANSPARENT_HUGEPAGE
836 default TRANSPARENT_HUGEPAGE_ALWAYS 783 default TRANSPARENT_HUGEPAGE_ALWAYS
837 help 784 help
838 Selects the sysfs defaults for Trans 785 Selects the sysfs defaults for Transparent Hugepage Support.
839 786
840 config TRANSPARENT_HUGEPAGE_ALWAYS 787 config TRANSPARENT_HUGEPAGE_ALWAYS
841 bool "always" 788 bool "always"
842 help 789 help
843 Enabling Transparent Hugepage always 790 Enabling Transparent Hugepage always, can increase the
844 memory footprint of applications wit 791 memory footprint of applications without a guaranteed
845 benefit but it will work automatical 792 benefit but it will work automatically for all applications.
846 793
847 config TRANSPARENT_HUGEPAGE_MADVISE 794 config TRANSPARENT_HUGEPAGE_MADVISE
848 bool "madvise" 795 bool "madvise"
849 help 796 help
850 Enabling Transparent Hugepage madvis 797 Enabling Transparent Hugepage madvise, will only provide a
851 performance improvement benefit to t 798 performance improvement benefit to the applications using
852 madvise(MADV_HUGEPAGE) but it won't 799 madvise(MADV_HUGEPAGE) but it won't risk to increase the
853 memory footprint of applications wit 800 memory footprint of applications without a guaranteed
854 benefit. 801 benefit.
855 <<
856 config TRANSPARENT_HUGEPAGE_NEVER <<
857 bool "never" <<
858 help <<
859 Disable Transparent Hugepage by defa <<
860 enabled at runtime via sysfs. <<
861 endchoice 802 endchoice
862 803
863 config THP_SWAP 804 config THP_SWAP
864 def_bool y 805 def_bool y
865 depends on TRANSPARENT_HUGEPAGE && ARC 806 depends on TRANSPARENT_HUGEPAGE && ARCH_WANTS_THP_SWAP && SWAP && 64BIT
866 help 807 help
867 Swap transparent huge pages in one p 808 Swap transparent huge pages in one piece, without splitting.
868 XXX: For now, swap cluster backing t 809 XXX: For now, swap cluster backing transparent huge page
869 will be split after swapout. 810 will be split after swapout.
870 811
871 For selection by architectures with 812 For selection by architectures with reasonable THP sizes.
872 813
873 config READ_ONLY_THP_FOR_FS 814 config READ_ONLY_THP_FOR_FS
874 bool "Read-only THP for filesystems (E 815 bool "Read-only THP for filesystems (EXPERIMENTAL)"
875 depends on TRANSPARENT_HUGEPAGE && SHM 816 depends on TRANSPARENT_HUGEPAGE && SHMEM
876 817
877 help 818 help
878 Allow khugepaged to put read-only fi 819 Allow khugepaged to put read-only file-backed pages in THP.
879 820
880 This is marked experimental because 821 This is marked experimental because it is a new feature. Write
881 support of file THPs will be develop 822 support of file THPs will be developed in the next few release
882 cycles. 823 cycles.
883 824
884 endif # TRANSPARENT_HUGEPAGE 825 endif # TRANSPARENT_HUGEPAGE
885 826
886 # 827 #
887 # The architecture supports pgtable leaves tha <<
888 # <<
889 config PGTABLE_HAS_HUGE_LEAVES <<
890 def_bool TRANSPARENT_HUGEPAGE || HUGET <<
891 <<
892 # TODO: Allow to be enabled without THP <<
893 config ARCH_SUPPORTS_HUGE_PFNMAP <<
894 def_bool n <<
895 depends on TRANSPARENT_HUGEPAGE <<
896 <<
897 config ARCH_SUPPORTS_PMD_PFNMAP <<
898 def_bool y <<
899 depends on ARCH_SUPPORTS_HUGE_PFNMAP & <<
900 <<
901 config ARCH_SUPPORTS_PUD_PFNMAP <<
902 def_bool y <<
903 depends on ARCH_SUPPORTS_HUGE_PFNMAP & <<
904 <<
905 # <<
906 # UP and nommu archs use km based percpu alloc 828 # UP and nommu archs use km based percpu allocator
907 # 829 #
908 config NEED_PER_CPU_KM 830 config NEED_PER_CPU_KM
909 depends on !SMP || !MMU 831 depends on !SMP || !MMU
910 bool 832 bool
911 default y 833 default y
912 834
913 config NEED_PER_CPU_EMBED_FIRST_CHUNK 835 config NEED_PER_CPU_EMBED_FIRST_CHUNK
914 bool 836 bool
915 837
916 config NEED_PER_CPU_PAGE_FIRST_CHUNK 838 config NEED_PER_CPU_PAGE_FIRST_CHUNK
917 bool 839 bool
918 840
919 config USE_PERCPU_NUMA_NODE_ID 841 config USE_PERCPU_NUMA_NODE_ID
920 bool 842 bool
921 843
922 config HAVE_SETUP_PER_CPU_AREA 844 config HAVE_SETUP_PER_CPU_AREA
923 bool 845 bool
924 846
>> 847 config FRONTSWAP
>> 848 bool
>> 849
925 config CMA 850 config CMA
926 bool "Contiguous Memory Allocator" 851 bool "Contiguous Memory Allocator"
927 depends on MMU 852 depends on MMU
928 select MIGRATION 853 select MIGRATION
929 select MEMORY_ISOLATION 854 select MEMORY_ISOLATION
930 help 855 help
931 This enables the Contiguous Memory A 856 This enables the Contiguous Memory Allocator which allows other
932 subsystems to allocate big physicall 857 subsystems to allocate big physically-contiguous blocks of memory.
933 CMA reserves a region of memory and 858 CMA reserves a region of memory and allows only movable pages to
934 be allocated from it. This way, the 859 be allocated from it. This way, the kernel can use the memory for
935 pagecache and when a subsystem reque 860 pagecache and when a subsystem requests for contiguous area, the
936 allocated pages are migrated away to 861 allocated pages are migrated away to serve the contiguous request.
937 862
938 If unsure, say "n". 863 If unsure, say "n".
939 864
>> 865 config CMA_DEBUG
>> 866 bool "CMA debug messages (DEVELOPMENT)"
>> 867 depends on DEBUG_KERNEL && CMA
>> 868 help
>> 869 Turns on debug messages in CMA. This produces KERN_DEBUG
>> 870 messages for every CMA call as well as various messages while
>> 871 processing calls such as dma_alloc_from_contiguous().
>> 872 This option does not affect warning and error messages.
>> 873
940 config CMA_DEBUGFS 874 config CMA_DEBUGFS
941 bool "CMA debugfs interface" 875 bool "CMA debugfs interface"
942 depends on CMA && DEBUG_FS 876 depends on CMA && DEBUG_FS
943 help 877 help
944 Turns on the DebugFS interface for C 878 Turns on the DebugFS interface for CMA.
945 879
946 config CMA_SYSFS 880 config CMA_SYSFS
947 bool "CMA information through sysfs in 881 bool "CMA information through sysfs interface"
948 depends on CMA && SYSFS 882 depends on CMA && SYSFS
949 help 883 help
950 This option exposes some sysfs attri 884 This option exposes some sysfs attributes to get information
951 from CMA. 885 from CMA.
952 886
953 config CMA_AREAS 887 config CMA_AREAS
954 int "Maximum count of the CMA areas" 888 int "Maximum count of the CMA areas"
955 depends on CMA 889 depends on CMA
956 default 20 if NUMA !! 890 default 19 if NUMA
957 default 8 !! 891 default 7
958 help 892 help
959 CMA allows to create CMA areas for p 893 CMA allows to create CMA areas for particular purpose, mainly,
960 used as device private area. This pa 894 used as device private area. This parameter sets the maximum
961 number of CMA area in the system. 895 number of CMA area in the system.
962 896
963 If unsure, leave the default value " !! 897 If unsure, leave the default value "7" in UMA and "19" in NUMA.
964 898
965 config MEM_SOFT_DIRTY 899 config MEM_SOFT_DIRTY
966 bool "Track memory changes" 900 bool "Track memory changes"
967 depends on CHECKPOINT_RESTORE && HAVE_ 901 depends on CHECKPOINT_RESTORE && HAVE_ARCH_SOFT_DIRTY && PROC_FS
968 select PROC_PAGE_MONITOR 902 select PROC_PAGE_MONITOR
969 help 903 help
970 This option enables memory changes t 904 This option enables memory changes tracking by introducing a
971 soft-dirty bit on pte-s. This bit it 905 soft-dirty bit on pte-s. This bit it set when someone writes
972 into a page just as regular dirty bi 906 into a page just as regular dirty bit, but unlike the latter
973 it can be cleared by hands. 907 it can be cleared by hands.
974 908
975 See Documentation/admin-guide/mm/sof 909 See Documentation/admin-guide/mm/soft-dirty.rst for more details.
976 910
977 config GENERIC_EARLY_IOREMAP 911 config GENERIC_EARLY_IOREMAP
978 bool 912 bool
979 913
980 config STACK_MAX_DEFAULT_SIZE_MB 914 config STACK_MAX_DEFAULT_SIZE_MB
981 int "Default maximum user stack size f 915 int "Default maximum user stack size for 32-bit processes (MB)"
982 default 100 916 default 100
983 range 8 2048 917 range 8 2048
984 depends on STACK_GROWSUP && (!64BIT || 918 depends on STACK_GROWSUP && (!64BIT || COMPAT)
985 help 919 help
986 This is the maximum stack size in Me 920 This is the maximum stack size in Megabytes in the VM layout of 32-bit
987 user processes when the stack grows 921 user processes when the stack grows upwards (currently only on parisc
988 arch) when the RLIMIT_STACK hard lim 922 arch) when the RLIMIT_STACK hard limit is unlimited.
989 923
990 A sane initial value is 100 MB. 924 A sane initial value is 100 MB.
991 925
992 config DEFERRED_STRUCT_PAGE_INIT 926 config DEFERRED_STRUCT_PAGE_INIT
993 bool "Defer initialisation of struct p 927 bool "Defer initialisation of struct pages to kthreads"
994 depends on SPARSEMEM 928 depends on SPARSEMEM
995 depends on !NEED_PER_CPU_KM 929 depends on !NEED_PER_CPU_KM
996 depends on 64BIT 930 depends on 64BIT
997 depends on !KMSAN <<
998 select PADATA 931 select PADATA
999 help 932 help
1000 Ordinarily all struct pages are ini 933 Ordinarily all struct pages are initialised during early boot in a
1001 single thread. On very large machin 934 single thread. On very large machines this can take a considerable
1002 amount of time. If this option is s 935 amount of time. If this option is set, large machines will bring up
1003 a subset of memmap at boot and then 936 a subset of memmap at boot and then initialise the rest in parallel.
1004 This has a potential performance im 937 This has a potential performance impact on tasks running early in the
1005 lifetime of the system until these 938 lifetime of the system until these kthreads finish the
1006 initialisation. 939 initialisation.
1007 940
1008 config PAGE_IDLE_FLAG 941 config PAGE_IDLE_FLAG
1009 bool 942 bool
1010 select PAGE_EXTENSION if !64BIT 943 select PAGE_EXTENSION if !64BIT
1011 help 944 help
1012 This adds PG_idle and PG_young flag 945 This adds PG_idle and PG_young flags to 'struct page'. PTE Accessed
1013 bit writers can set the state of th 946 bit writers can set the state of the bit in the flags so that PTE
1014 Accessed bit readers may avoid dist 947 Accessed bit readers may avoid disturbance.
1015 948
1016 config IDLE_PAGE_TRACKING 949 config IDLE_PAGE_TRACKING
1017 bool "Enable idle page tracking" 950 bool "Enable idle page tracking"
1018 depends on SYSFS && MMU 951 depends on SYSFS && MMU
1019 select PAGE_IDLE_FLAG 952 select PAGE_IDLE_FLAG
1020 help 953 help
1021 This feature allows to estimate the 954 This feature allows to estimate the amount of user pages that have
1022 not been touched during a given per 955 not been touched during a given period of time. This information can
1023 be useful to tune memory cgroup lim 956 be useful to tune memory cgroup limits and/or for job placement
1024 within a compute cluster. 957 within a compute cluster.
1025 958
1026 See Documentation/admin-guide/mm/id 959 See Documentation/admin-guide/mm/idle_page_tracking.rst for
1027 more details. 960 more details.
1028 961
1029 # Architectures which implement cpu_dcache_is <<
1030 # whether the data caches are aliased (VIVT o <<
1031 # aliasing) need to select this. <<
1032 config ARCH_HAS_CPU_CACHE_ALIASING <<
1033 bool <<
1034 <<
1035 config ARCH_HAS_CACHE_LINE_SIZE 962 config ARCH_HAS_CACHE_LINE_SIZE
1036 bool 963 bool
1037 964
1038 config ARCH_HAS_CURRENT_STACK_POINTER 965 config ARCH_HAS_CURRENT_STACK_POINTER
1039 bool 966 bool
1040 help 967 help
1041 In support of HARDENED_USERCOPY per 968 In support of HARDENED_USERCOPY performing stack variable lifetime
1042 checking, an architecture-agnostic 969 checking, an architecture-agnostic way to find the stack pointer
1043 is needed. Once an architecture def 970 is needed. Once an architecture defines an unsigned long global
1044 register alias named "current_stack 971 register alias named "current_stack_pointer", this config can be
1045 selected. 972 selected.
1046 973
1047 config ARCH_HAS_PTE_DEVMAP 974 config ARCH_HAS_PTE_DEVMAP
1048 bool 975 bool
1049 976
1050 config ARCH_HAS_ZONE_DMA_SET 977 config ARCH_HAS_ZONE_DMA_SET
1051 bool 978 bool
1052 979
1053 config ZONE_DMA 980 config ZONE_DMA
1054 bool "Support DMA zone" if ARCH_HAS_Z 981 bool "Support DMA zone" if ARCH_HAS_ZONE_DMA_SET
1055 default y if ARM64 || X86 982 default y if ARM64 || X86
1056 983
1057 config ZONE_DMA32 984 config ZONE_DMA32
1058 bool "Support DMA32 zone" if ARCH_HAS 985 bool "Support DMA32 zone" if ARCH_HAS_ZONE_DMA_SET
1059 depends on !X86_32 986 depends on !X86_32
1060 default y if ARM64 987 default y if ARM64
1061 988
1062 config ZONE_DEVICE 989 config ZONE_DEVICE
1063 bool "Device memory (pmem, HMM, etc.. 990 bool "Device memory (pmem, HMM, etc...) hotplug support"
1064 depends on MEMORY_HOTPLUG 991 depends on MEMORY_HOTPLUG
1065 depends on MEMORY_HOTREMOVE 992 depends on MEMORY_HOTREMOVE
1066 depends on SPARSEMEM_VMEMMAP 993 depends on SPARSEMEM_VMEMMAP
1067 depends on ARCH_HAS_PTE_DEVMAP 994 depends on ARCH_HAS_PTE_DEVMAP
1068 select XARRAY_MULTI 995 select XARRAY_MULTI
1069 996
1070 help 997 help
1071 Device memory hotplug support allow 998 Device memory hotplug support allows for establishing pmem,
1072 or other device driver discovered m 999 or other device driver discovered memory regions, in the
1073 memmap. This allows pfn_to_page() l 1000 memmap. This allows pfn_to_page() lookups of otherwise
1074 "device-physical" addresses which i 1001 "device-physical" addresses which is needed for using a DAX
1075 mapping in an O_DIRECT operation, a 1002 mapping in an O_DIRECT operation, among other things.
1076 1003
1077 If FS_DAX is enabled, then say Y. 1004 If FS_DAX is enabled, then say Y.
1078 1005
1079 # 1006 #
1080 # Helpers to mirror range of the CPU page tab 1007 # Helpers to mirror range of the CPU page tables of a process into device page
1081 # tables. 1008 # tables.
1082 # 1009 #
1083 config HMM_MIRROR 1010 config HMM_MIRROR
1084 bool 1011 bool
1085 depends on MMU 1012 depends on MMU
1086 1013
1087 config GET_FREE_REGION 1014 config GET_FREE_REGION
>> 1015 depends on SPARSEMEM
1088 bool 1016 bool
1089 1017
1090 config DEVICE_PRIVATE 1018 config DEVICE_PRIVATE
1091 bool "Unaddressable device memory (GP 1019 bool "Unaddressable device memory (GPU memory, ...)"
1092 depends on ZONE_DEVICE 1020 depends on ZONE_DEVICE
1093 select GET_FREE_REGION 1021 select GET_FREE_REGION
1094 1022
1095 help 1023 help
1096 Allows creation of struct pages to 1024 Allows creation of struct pages to represent unaddressable device
1097 memory; i.e., memory that is only a 1025 memory; i.e., memory that is only accessible from the device (or
1098 group of devices). You likely also 1026 group of devices). You likely also want to select HMM_MIRROR.
1099 1027
1100 config VMAP_PFN 1028 config VMAP_PFN
1101 bool 1029 bool
1102 1030
1103 config ARCH_USES_HIGH_VMA_FLAGS 1031 config ARCH_USES_HIGH_VMA_FLAGS
1104 bool 1032 bool
1105 config ARCH_HAS_PKEYS 1033 config ARCH_HAS_PKEYS
1106 bool 1034 bool
1107 1035
1108 config ARCH_USES_PG_ARCH_2 !! 1036 config ARCH_USES_PG_ARCH_X
1109 bool <<
1110 config ARCH_USES_PG_ARCH_3 <<
1111 bool 1037 bool
>> 1038 help
>> 1039 Enable the definition of PG_arch_x page flags with x > 1. Only
>> 1040 suitable for 64-bit architectures with CONFIG_FLATMEM or
>> 1041 CONFIG_SPARSEMEM_VMEMMAP enabled, otherwise there may not be
>> 1042 enough room for additional bits in page->flags.
1112 1043
1113 config VM_EVENT_COUNTERS 1044 config VM_EVENT_COUNTERS
1114 default y 1045 default y
1115 bool "Enable VM event counters for /p 1046 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1116 help 1047 help
1117 VM event counters are needed for ev 1048 VM event counters are needed for event counts to be shown.
1118 This option allows the disabling of 1049 This option allows the disabling of the VM event counters
1119 on EXPERT systems. /proc/vmstat wi 1050 on EXPERT systems. /proc/vmstat will only show page counts
1120 if VM event counters are disabled. 1051 if VM event counters are disabled.
1121 1052
1122 config PERCPU_STATS 1053 config PERCPU_STATS
1123 bool "Collect percpu memory statistic 1054 bool "Collect percpu memory statistics"
1124 help 1055 help
1125 This feature collects and exposes s 1056 This feature collects and exposes statistics via debugfs. The
1126 information includes global and per 1057 information includes global and per chunk statistics, which can
1127 be used to help understand percpu m 1058 be used to help understand percpu memory usage.
1128 1059
1129 config GUP_TEST 1060 config GUP_TEST
1130 bool "Enable infrastructure for get_u 1061 bool "Enable infrastructure for get_user_pages()-related unit tests"
1131 depends on DEBUG_FS 1062 depends on DEBUG_FS
1132 help 1063 help
1133 Provides /sys/kernel/debug/gup_test 1064 Provides /sys/kernel/debug/gup_test, which in turn provides a way
1134 to make ioctl calls that can launch 1065 to make ioctl calls that can launch kernel-based unit tests for
1135 the get_user_pages*() and pin_user_ 1066 the get_user_pages*() and pin_user_pages*() family of API calls.
1136 1067
1137 These tests include benchmark testi 1068 These tests include benchmark testing of the _fast variants of
1138 get_user_pages*() and pin_user_page 1069 get_user_pages*() and pin_user_pages*(), as well as smoke tests of
1139 the non-_fast variants. 1070 the non-_fast variants.
1140 1071
1141 There is also a sub-test that allow 1072 There is also a sub-test that allows running dump_page() on any
1142 of up to eight pages (selected by c 1073 of up to eight pages (selected by command line args) within the
1143 range of user-space addresses. Thes 1074 range of user-space addresses. These pages are either pinned via
1144 pin_user_pages*(), or pinned via ge 1075 pin_user_pages*(), or pinned via get_user_pages*(), as specified
1145 by other command line arguments. 1076 by other command line arguments.
1146 1077
1147 See tools/testing/selftests/mm/gup_ 1078 See tools/testing/selftests/mm/gup_test.c
1148 1079
1149 comment "GUP_TEST needs to have DEBUG_FS enab 1080 comment "GUP_TEST needs to have DEBUG_FS enabled"
1150 depends on !GUP_TEST && !DEBUG_FS 1081 depends on !GUP_TEST && !DEBUG_FS
1151 1082
1152 config GUP_GET_PXX_LOW_HIGH 1083 config GUP_GET_PXX_LOW_HIGH
1153 bool 1084 bool
1154 1085
1155 config DMAPOOL_TEST 1086 config DMAPOOL_TEST
1156 tristate "Enable a module to run time 1087 tristate "Enable a module to run time tests on dma_pool"
1157 depends on HAS_DMA 1088 depends on HAS_DMA
1158 help 1089 help
1159 Provides a test module that will al 1090 Provides a test module that will allocate and free many blocks of
1160 various sizes and report how long i 1091 various sizes and report how long it takes. This is intended to
1161 provide a consistent way to measure 1092 provide a consistent way to measure how changes to the
1162 dma_pool_alloc/free routines affect 1093 dma_pool_alloc/free routines affect performance.
1163 1094
1164 config ARCH_HAS_PTE_SPECIAL 1095 config ARCH_HAS_PTE_SPECIAL
1165 bool 1096 bool
1166 1097
>> 1098 #
>> 1099 # Some architectures require a special hugepage directory format that is
>> 1100 # required to support multiple hugepage sizes. For example a4fe3ce76
>> 1101 # "powerpc/mm: Allow more flexible layouts for hugepage pagetables"
>> 1102 # introduced it on powerpc. This allows for a more flexible hugepage
>> 1103 # pagetable layouts.
>> 1104 #
>> 1105 config ARCH_HAS_HUGEPD
>> 1106 bool
>> 1107
1167 config MAPPING_DIRTY_HELPERS 1108 config MAPPING_DIRTY_HELPERS
1168 bool 1109 bool
1169 1110
1170 config KMAP_LOCAL 1111 config KMAP_LOCAL
1171 bool 1112 bool
1172 1113
1173 config KMAP_LOCAL_NON_LINEAR_PTE_ARRAY 1114 config KMAP_LOCAL_NON_LINEAR_PTE_ARRAY
1174 bool 1115 bool
1175 1116
1176 # struct io_mapping based helper. Selected b 1117 # struct io_mapping based helper. Selected by drivers that need them
1177 config IO_MAPPING 1118 config IO_MAPPING
1178 bool 1119 bool
1179 1120
1180 config MEMFD_CREATE <<
1181 bool "Enable memfd_create() system ca <<
1182 <<
1183 config SECRETMEM 1121 config SECRETMEM
1184 default y 1122 default y
1185 bool "Enable memfd_secret() system ca 1123 bool "Enable memfd_secret() system call" if EXPERT
1186 depends on ARCH_HAS_SET_DIRECT_MAP 1124 depends on ARCH_HAS_SET_DIRECT_MAP
1187 help 1125 help
1188 Enable the memfd_secret() system ca 1126 Enable the memfd_secret() system call with the ability to create
1189 memory areas visible only in the co 1127 memory areas visible only in the context of the owning process and
1190 not mapped to other processes and o 1128 not mapped to other processes and other kernel page tables.
1191 1129
1192 config ANON_VMA_NAME 1130 config ANON_VMA_NAME
1193 bool "Anonymous VMA name support" 1131 bool "Anonymous VMA name support"
1194 depends on PROC_FS && ADVISE_SYSCALLS 1132 depends on PROC_FS && ADVISE_SYSCALLS && MMU
1195 1133
1196 help 1134 help
1197 Allow naming anonymous virtual memo 1135 Allow naming anonymous virtual memory areas.
1198 1136
1199 This feature allows assigning names 1137 This feature allows assigning names to virtual memory areas. Assigned
1200 names can be later retrieved from / 1138 names can be later retrieved from /proc/pid/maps and /proc/pid/smaps
1201 and help identifying individual ano 1139 and help identifying individual anonymous memory areas.
1202 Assigning a name to anonymous virtu 1140 Assigning a name to anonymous virtual memory area might prevent that
1203 area from being merged with adjacen 1141 area from being merged with adjacent virtual memory areas due to the
1204 difference in their name. 1142 difference in their name.
1205 1143
>> 1144 config USERFAULTFD
>> 1145 bool "Enable userfaultfd() system call"
>> 1146 depends on MMU
>> 1147 help
>> 1148 Enable the userfaultfd() system call that allows to intercept and
>> 1149 handle page faults in userland.
>> 1150
1206 config HAVE_ARCH_USERFAULTFD_WP 1151 config HAVE_ARCH_USERFAULTFD_WP
1207 bool 1152 bool
1208 help 1153 help
1209 Arch has userfaultfd write protecti 1154 Arch has userfaultfd write protection support
1210 1155
1211 config HAVE_ARCH_USERFAULTFD_MINOR 1156 config HAVE_ARCH_USERFAULTFD_MINOR
1212 bool 1157 bool
1213 help 1158 help
1214 Arch has userfaultfd minor fault su 1159 Arch has userfaultfd minor fault support
1215 1160
1216 menuconfig USERFAULTFD <<
1217 bool "Enable userfaultfd() system cal <<
1218 depends on MMU <<
1219 help <<
1220 Enable the userfaultfd() system cal <<
1221 handle page faults in userland. <<
1222 <<
1223 if USERFAULTFD <<
1224 config PTE_MARKER_UFFD_WP 1161 config PTE_MARKER_UFFD_WP
1225 bool "Userfaultfd write protection su 1162 bool "Userfaultfd write protection support for shmem/hugetlbfs"
1226 default y 1163 default y
1227 depends on HAVE_ARCH_USERFAULTFD_WP 1164 depends on HAVE_ARCH_USERFAULTFD_WP
1228 1165
1229 help 1166 help
1230 Allows to create marker PTEs for us 1167 Allows to create marker PTEs for userfaultfd write protection
1231 purposes. It is required to enable 1168 purposes. It is required to enable userfaultfd write protection on
1232 file-backed memory types like shmem 1169 file-backed memory types like shmem and hugetlbfs.
1233 endif # USERFAULTFD <<
1234 1170
1235 # multi-gen LRU { 1171 # multi-gen LRU {
1236 config LRU_GEN 1172 config LRU_GEN
1237 bool "Multi-Gen LRU" 1173 bool "Multi-Gen LRU"
1238 depends on MMU 1174 depends on MMU
1239 # make sure folio->flags has enough s 1175 # make sure folio->flags has enough spare bits
1240 depends on 64BIT || !SPARSEMEM || SPA 1176 depends on 64BIT || !SPARSEMEM || SPARSEMEM_VMEMMAP
1241 help 1177 help
1242 A high performance LRU implementati 1178 A high performance LRU implementation to overcommit memory. See
1243 Documentation/admin-guide/mm/multig 1179 Documentation/admin-guide/mm/multigen_lru.rst for details.
1244 1180
1245 config LRU_GEN_ENABLED 1181 config LRU_GEN_ENABLED
1246 bool "Enable by default" 1182 bool "Enable by default"
1247 depends on LRU_GEN 1183 depends on LRU_GEN
1248 help 1184 help
1249 This option enables the multi-gen L 1185 This option enables the multi-gen LRU by default.
1250 1186
1251 config LRU_GEN_STATS 1187 config LRU_GEN_STATS
1252 bool "Full stats for debugging" 1188 bool "Full stats for debugging"
1253 depends on LRU_GEN 1189 depends on LRU_GEN
1254 help 1190 help
1255 Do not enable this option unless yo 1191 Do not enable this option unless you plan to look at historical stats
1256 from evicted generations for debugg 1192 from evicted generations for debugging purpose.
1257 1193
1258 This option has a per-memcg and per 1194 This option has a per-memcg and per-node memory overhead.
1259 <<
1260 config LRU_GEN_WALKS_MMU <<
1261 def_bool y <<
1262 depends on LRU_GEN && ARCH_HAS_HW_PTE <<
1263 # } 1195 # }
1264 1196
1265 config ARCH_SUPPORTS_PER_VMA_LOCK 1197 config ARCH_SUPPORTS_PER_VMA_LOCK
1266 def_bool n 1198 def_bool n
1267 1199
1268 config PER_VMA_LOCK 1200 config PER_VMA_LOCK
1269 def_bool y 1201 def_bool y
1270 depends on ARCH_SUPPORTS_PER_VMA_LOCK 1202 depends on ARCH_SUPPORTS_PER_VMA_LOCK && MMU && SMP
1271 help 1203 help
1272 Allow per-vma locking during page f 1204 Allow per-vma locking during page fault handling.
1273 1205
1274 This feature allows locking each vi 1206 This feature allows locking each virtual memory area separately when
1275 handling page faults instead of tak 1207 handling page faults instead of taking mmap_lock.
1276 1208
1277 config LOCK_MM_AND_FIND_VMA 1209 config LOCK_MM_AND_FIND_VMA
1278 bool 1210 bool
1279 depends on !STACK_GROWSUP 1211 depends on !STACK_GROWSUP
1280 <<
1281 config IOMMU_MM_DATA <<
1282 bool <<
1283 <<
1284 config EXECMEM <<
1285 bool <<
1286 <<
1287 config NUMA_MEMBLKS <<
1288 bool <<
1289 <<
1290 config NUMA_EMU <<
1291 bool "NUMA emulation" <<
1292 depends on NUMA_MEMBLKS <<
1293 help <<
1294 Enable NUMA emulation. A flat machi <<
1295 into virtual nodes when booted with <<
1296 number of nodes. This is only usefu <<
1297 1212
1298 source "mm/damon/Kconfig" 1213 source "mm/damon/Kconfig"
1299 1214
1300 endmenu 1215 endmenu
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.