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