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