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