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