1 =============================== 1 ===============================
2 Documentation for /proc/sys/vm/ 2 Documentation for /proc/sys/vm/
3 =============================== 3 ===============================
4 4
5 kernel version 2.6.29 5 kernel version 2.6.29
6 6
7 Copyright (c) 1998, 1999, Rik van Riel <riel@n 7 Copyright (c) 1998, 1999, Rik van Riel <riel@nl.linux.org>
8 8
9 Copyright (c) 2008 Peter W. Morreale <p 9 Copyright (c) 2008 Peter W. Morreale <pmorreale@novell.com>
10 10
11 For general info and legal blurb, please look 11 For general info and legal blurb, please look in index.rst.
12 12
13 ---------------------------------------------- 13 ------------------------------------------------------------------------------
14 14
15 This file contains the documentation for the s 15 This file contains the documentation for the sysctl files in
16 /proc/sys/vm and is valid for Linux kernel ver 16 /proc/sys/vm and is valid for Linux kernel version 2.6.29.
17 17
18 The files in this directory can be used to tun 18 The files in this directory can be used to tune the operation
19 of the virtual memory (VM) subsystem of the Li 19 of the virtual memory (VM) subsystem of the Linux kernel and
20 the writeout of dirty data to disk. 20 the writeout of dirty data to disk.
21 21
22 Default values and initialization routines for 22 Default values and initialization routines for most of these
23 files can be found in mm/swap.c. 23 files can be found in mm/swap.c.
24 24
25 Currently, these files are in /proc/sys/vm: 25 Currently, these files are in /proc/sys/vm:
26 26
27 - admin_reserve_kbytes 27 - admin_reserve_kbytes
28 - compact_memory 28 - compact_memory
29 - compaction_proactiveness 29 - compaction_proactiveness
30 - compact_unevictable_allowed 30 - compact_unevictable_allowed
31 - dirty_background_bytes 31 - dirty_background_bytes
32 - dirty_background_ratio 32 - dirty_background_ratio
33 - dirty_bytes 33 - dirty_bytes
34 - dirty_expire_centisecs 34 - dirty_expire_centisecs
35 - dirty_ratio 35 - dirty_ratio
36 - dirtytime_expire_seconds 36 - dirtytime_expire_seconds
37 - dirty_writeback_centisecs 37 - dirty_writeback_centisecs
38 - drop_caches 38 - drop_caches
39 - enable_soft_offline <<
40 - extfrag_threshold 39 - extfrag_threshold
41 - highmem_is_dirtyable 40 - highmem_is_dirtyable
42 - hugetlb_shm_group 41 - hugetlb_shm_group
43 - laptop_mode 42 - laptop_mode
44 - legacy_va_layout 43 - legacy_va_layout
45 - lowmem_reserve_ratio 44 - lowmem_reserve_ratio
46 - max_map_count 45 - max_map_count
47 - mem_profiling (only if CONFIG_MEM_AL 46 - mem_profiling (only if CONFIG_MEM_ALLOC_PROFILING=y)
48 - memory_failure_early_kill 47 - memory_failure_early_kill
49 - memory_failure_recovery 48 - memory_failure_recovery
50 - min_free_kbytes 49 - min_free_kbytes
51 - min_slab_ratio 50 - min_slab_ratio
52 - min_unmapped_ratio 51 - min_unmapped_ratio
53 - mmap_min_addr 52 - mmap_min_addr
54 - mmap_rnd_bits 53 - mmap_rnd_bits
55 - mmap_rnd_compat_bits 54 - mmap_rnd_compat_bits
56 - nr_hugepages 55 - nr_hugepages
57 - nr_hugepages_mempolicy 56 - nr_hugepages_mempolicy
58 - nr_overcommit_hugepages 57 - nr_overcommit_hugepages
59 - nr_trim_pages (only if CONFIG_MMU=n) 58 - nr_trim_pages (only if CONFIG_MMU=n)
60 - numa_zonelist_order 59 - numa_zonelist_order
61 - oom_dump_tasks 60 - oom_dump_tasks
62 - oom_kill_allocating_task 61 - oom_kill_allocating_task
63 - overcommit_kbytes 62 - overcommit_kbytes
64 - overcommit_memory 63 - overcommit_memory
65 - overcommit_ratio 64 - overcommit_ratio
66 - page-cluster 65 - page-cluster
67 - page_lock_unfairness 66 - page_lock_unfairness
68 - panic_on_oom 67 - panic_on_oom
69 - percpu_pagelist_high_fraction 68 - percpu_pagelist_high_fraction
70 - stat_interval 69 - stat_interval
71 - stat_refresh 70 - stat_refresh
72 - numa_stat 71 - numa_stat
73 - swappiness 72 - swappiness
74 - unprivileged_userfaultfd 73 - unprivileged_userfaultfd
75 - user_reserve_kbytes 74 - user_reserve_kbytes
76 - vfs_cache_pressure 75 - vfs_cache_pressure
77 - watermark_boost_factor 76 - watermark_boost_factor
78 - watermark_scale_factor 77 - watermark_scale_factor
79 - zone_reclaim_mode 78 - zone_reclaim_mode
80 79
81 80
82 admin_reserve_kbytes 81 admin_reserve_kbytes
83 ==================== 82 ====================
84 83
85 The amount of free memory in the system that s 84 The amount of free memory in the system that should be reserved for users
86 with the capability cap_sys_admin. 85 with the capability cap_sys_admin.
87 86
88 admin_reserve_kbytes defaults to min(3% of fre 87 admin_reserve_kbytes defaults to min(3% of free pages, 8MB)
89 88
90 That should provide enough for the admin to lo 89 That should provide enough for the admin to log in and kill a process,
91 if necessary, under the default overcommit 'gu 90 if necessary, under the default overcommit 'guess' mode.
92 91
93 Systems running under overcommit 'never' shoul 92 Systems running under overcommit 'never' should increase this to account
94 for the full Virtual Memory Size of programs u 93 for the full Virtual Memory Size of programs used to recover. Otherwise,
95 root may not be able to log in to recover the 94 root may not be able to log in to recover the system.
96 95
97 How do you calculate a minimum useful reserve? 96 How do you calculate a minimum useful reserve?
98 97
99 sshd or login + bash (or some other shell) + t 98 sshd or login + bash (or some other shell) + top (or ps, kill, etc.)
100 99
101 For overcommit 'guess', we can sum resident se 100 For overcommit 'guess', we can sum resident set sizes (RSS).
102 On x86_64 this is about 8MB. 101 On x86_64 this is about 8MB.
103 102
104 For overcommit 'never', we can take the max of 103 For overcommit 'never', we can take the max of their virtual sizes (VSZ)
105 and add the sum of their RSS. 104 and add the sum of their RSS.
106 On x86_64 this is about 128MB. 105 On x86_64 this is about 128MB.
107 106
108 Changing this takes effect whenever an applica 107 Changing this takes effect whenever an application requests memory.
109 108
110 109
111 compact_memory 110 compact_memory
112 ============== 111 ==============
113 112
114 Available only when CONFIG_COMPACTION is set. 113 Available only when CONFIG_COMPACTION is set. When 1 is written to the file,
115 all zones are compacted such that free memory 114 all zones are compacted such that free memory is available in contiguous
116 blocks where possible. This can be important f 115 blocks where possible. This can be important for example in the allocation of
117 huge pages although processes will also direct 116 huge pages although processes will also directly compact memory as required.
118 117
119 compaction_proactiveness 118 compaction_proactiveness
120 ======================== 119 ========================
121 120
122 This tunable takes a value in the range [0, 10 121 This tunable takes a value in the range [0, 100] with a default value of
123 20. This tunable determines how aggressively c 122 20. This tunable determines how aggressively compaction is done in the
124 background. Write of a non zero value to this 123 background. Write of a non zero value to this tunable will immediately
125 trigger the proactive compaction. Setting it t 124 trigger the proactive compaction. Setting it to 0 disables proactive compaction.
126 125
127 Note that compaction has a non-trivial system- 126 Note that compaction has a non-trivial system-wide impact as pages
128 belonging to different processes are moved aro 127 belonging to different processes are moved around, which could also lead
129 to latency spikes in unsuspecting applications 128 to latency spikes in unsuspecting applications. The kernel employs
130 various heuristics to avoid wasting CPU cycles 129 various heuristics to avoid wasting CPU cycles if it detects that
131 proactive compaction is not being effective. 130 proactive compaction is not being effective.
132 131
133 Be careful when setting it to extreme values l 132 Be careful when setting it to extreme values like 100, as that may
134 cause excessive background compaction activity 133 cause excessive background compaction activity.
135 134
136 compact_unevictable_allowed 135 compact_unevictable_allowed
137 =========================== 136 ===========================
138 137
139 Available only when CONFIG_COMPACTION is set. 138 Available only when CONFIG_COMPACTION is set. When set to 1, compaction is
140 allowed to examine the unevictable lru (mlocke 139 allowed to examine the unevictable lru (mlocked pages) for pages to compact.
141 This should be used on systems where stalls fo 140 This should be used on systems where stalls for minor page faults are an
142 acceptable trade for large contiguous free mem 141 acceptable trade for large contiguous free memory. Set to 0 to prevent
143 compaction from moving pages that are unevicta 142 compaction from moving pages that are unevictable. Default value is 1.
144 On CONFIG_PREEMPT_RT the default value is 0 in 143 On CONFIG_PREEMPT_RT the default value is 0 in order to avoid a page fault, due
145 to compaction, which would block the task from 144 to compaction, which would block the task from becoming active until the fault
146 is resolved. 145 is resolved.
147 146
148 147
149 dirty_background_bytes 148 dirty_background_bytes
150 ====================== 149 ======================
151 150
152 Contains the amount of dirty memory at which t 151 Contains the amount of dirty memory at which the background kernel
153 flusher threads will start writeback. 152 flusher threads will start writeback.
154 153
155 Note: 154 Note:
156 dirty_background_bytes is the counterpart of 155 dirty_background_bytes is the counterpart of dirty_background_ratio. Only
157 one of them may be specified at a time. When 156 one of them may be specified at a time. When one sysctl is written it is
158 immediately taken into account to evaluate t 157 immediately taken into account to evaluate the dirty memory limits and the
159 other appears as 0 when read. 158 other appears as 0 when read.
160 159
161 160
162 dirty_background_ratio 161 dirty_background_ratio
163 ====================== 162 ======================
164 163
165 Contains, as a percentage of total available m 164 Contains, as a percentage of total available memory that contains free pages
166 and reclaimable pages, the number of pages at 165 and reclaimable pages, the number of pages at which the background kernel
167 flusher threads will start writing out dirty d 166 flusher threads will start writing out dirty data.
168 167
169 The total available memory is not equal to tot 168 The total available memory is not equal to total system memory.
170 169
171 170
172 dirty_bytes 171 dirty_bytes
173 =========== 172 ===========
174 173
175 Contains the amount of dirty memory at which a 174 Contains the amount of dirty memory at which a process generating disk writes
176 will itself start writeback. 175 will itself start writeback.
177 176
178 Note: dirty_bytes is the counterpart of dirty_ 177 Note: dirty_bytes is the counterpart of dirty_ratio. Only one of them may be
179 specified at a time. When one sysctl is writte 178 specified at a time. When one sysctl is written it is immediately taken into
180 account to evaluate the dirty memory limits an 179 account to evaluate the dirty memory limits and the other appears as 0 when
181 read. 180 read.
182 181
183 Note: the minimum value allowed for dirty_byte 182 Note: the minimum value allowed for dirty_bytes is two pages (in bytes); any
184 value lower than this limit will be ignored an 183 value lower than this limit will be ignored and the old configuration will be
185 retained. 184 retained.
186 185
187 186
188 dirty_expire_centisecs 187 dirty_expire_centisecs
189 ====================== 188 ======================
190 189
191 This tunable is used to define when dirty data 190 This tunable is used to define when dirty data is old enough to be eligible
192 for writeout by the kernel flusher threads. I 191 for writeout by the kernel flusher threads. It is expressed in 100'ths
193 of a second. Data which has been dirty in-mem 192 of a second. Data which has been dirty in-memory for longer than this
194 interval will be written out next time a flush 193 interval will be written out next time a flusher thread wakes up.
195 194
196 195
197 dirty_ratio 196 dirty_ratio
198 =========== 197 ===========
199 198
200 Contains, as a percentage of total available m 199 Contains, as a percentage of total available memory that contains free pages
201 and reclaimable pages, the number of pages at 200 and reclaimable pages, the number of pages at which a process which is
202 generating disk writes will itself start writi 201 generating disk writes will itself start writing out dirty data.
203 202
204 The total available memory is not equal to tot 203 The total available memory is not equal to total system memory.
205 204
206 205
207 dirtytime_expire_seconds 206 dirtytime_expire_seconds
208 ======================== 207 ========================
209 208
210 When a lazytime inode is constantly having its 209 When a lazytime inode is constantly having its pages dirtied, the inode with
211 an updated timestamp will never get chance to 210 an updated timestamp will never get chance to be written out. And, if the
212 only thing that has happened on the file syste 211 only thing that has happened on the file system is a dirtytime inode caused
213 by an atime update, a worker will be scheduled 212 by an atime update, a worker will be scheduled to make sure that inode
214 eventually gets pushed out to disk. This tuna 213 eventually gets pushed out to disk. This tunable is used to define when dirty
215 inode is old enough to be eligible for writeba 214 inode is old enough to be eligible for writeback by the kernel flusher threads.
216 And, it is also used as the interval to wakeup 215 And, it is also used as the interval to wakeup dirtytime_writeback thread.
217 216
218 217
219 dirty_writeback_centisecs 218 dirty_writeback_centisecs
220 ========================= 219 =========================
221 220
222 The kernel flusher threads will periodically w 221 The kernel flusher threads will periodically wake up and write `old` data
223 out to disk. This tunable expresses the inter 222 out to disk. This tunable expresses the interval between those wakeups, in
224 100'ths of a second. 223 100'ths of a second.
225 224
226 Setting this to zero disables periodic writeba 225 Setting this to zero disables periodic writeback altogether.
227 226
228 227
229 drop_caches 228 drop_caches
230 =========== 229 ===========
231 230
232 Writing to this will cause the kernel to drop 231 Writing to this will cause the kernel to drop clean caches, as well as
233 reclaimable slab objects like dentries and ino 232 reclaimable slab objects like dentries and inodes. Once dropped, their
234 memory becomes free. 233 memory becomes free.
235 234
236 To free pagecache:: 235 To free pagecache::
237 236
238 echo 1 > /proc/sys/vm/drop_caches 237 echo 1 > /proc/sys/vm/drop_caches
239 238
240 To free reclaimable slab objects (includes den 239 To free reclaimable slab objects (includes dentries and inodes)::
241 240
242 echo 2 > /proc/sys/vm/drop_caches 241 echo 2 > /proc/sys/vm/drop_caches
243 242
244 To free slab objects and pagecache:: 243 To free slab objects and pagecache::
245 244
246 echo 3 > /proc/sys/vm/drop_caches 245 echo 3 > /proc/sys/vm/drop_caches
247 246
248 This is a non-destructive operation and will n 247 This is a non-destructive operation and will not free any dirty objects.
249 To increase the number of objects freed by thi 248 To increase the number of objects freed by this operation, the user may run
250 `sync` prior to writing to /proc/sys/vm/drop_c 249 `sync` prior to writing to /proc/sys/vm/drop_caches. This will minimize the
251 number of dirty objects on the system and crea 250 number of dirty objects on the system and create more candidates to be
252 dropped. 251 dropped.
253 252
254 This file is not a means to control the growth 253 This file is not a means to control the growth of the various kernel caches
255 (inodes, dentries, pagecache, etc...) These o 254 (inodes, dentries, pagecache, etc...) These objects are automatically
256 reclaimed by the kernel when memory is needed 255 reclaimed by the kernel when memory is needed elsewhere on the system.
257 256
258 Use of this file can cause performance problem 257 Use of this file can cause performance problems. Since it discards cached
259 objects, it may cost a significant amount of I 258 objects, it may cost a significant amount of I/O and CPU to recreate the
260 dropped objects, especially if they were under 259 dropped objects, especially if they were under heavy use. Because of this,
261 use outside of a testing or debugging environm 260 use outside of a testing or debugging environment is not recommended.
262 261
263 You may see informational messages in your ker 262 You may see informational messages in your kernel log when this file is
264 used:: 263 used::
265 264
266 cat (1234): drop_caches: 3 265 cat (1234): drop_caches: 3
267 266
268 These are informational only. They do not mea 267 These are informational only. They do not mean that anything is wrong
269 with your system. To disable them, echo 4 (bi 268 with your system. To disable them, echo 4 (bit 2) into drop_caches.
270 269
271 enable_soft_offline <<
272 =================== <<
273 Correctable memory errors are very common on s <<
274 solution for memory pages having (excessive) c <<
275 <<
276 For different types of page, soft-offline has <<
277 <<
278 - For a raw error page, soft-offline migrates <<
279 a new raw page. <<
280 <<
281 - For a page that is part of a transparent hug <<
282 transparent hugepage into raw pages, then mi <<
283 As a result, user is transparently backed by <<
284 memory access performance. <<
285 <<
286 - For a page that is part of a HugeTLB hugepag <<
287 the entire HugeTLB hugepage, during which a <<
288 as migration target. Then the original huge <<
289 pages without compensation, reducing the cap <<
290 <<
291 It is user's call to choose between reliabilit <<
292 physical memory) vs performance / capacity imp <<
293 HugeTLB cases. <<
294 <<
295 For all architectures, enable_soft_offline con <<
296 memory pages. When set to 1, kernel attempts <<
297 whenever it thinks needed. When set to 0, ker <<
298 the request to soft offline the pages. Its de <<
299 <<
300 It is worth mentioning that after setting enab <<
301 following requests to soft offline pages will <<
302 <<
303 - Request to soft offline pages from RAS Corre <<
304 <<
305 - On ARM, the request to soft offline pages fr <<
306 <<
307 - On PARISC, the request to soft offline pages <<
308 270
309 extfrag_threshold 271 extfrag_threshold
310 ================= 272 =================
311 273
312 This parameter affects whether the kernel will 274 This parameter affects whether the kernel will compact memory or direct
313 reclaim to satisfy a high-order allocation. Th 275 reclaim to satisfy a high-order allocation. The extfrag/extfrag_index file in
314 debugfs shows what the fragmentation index for 276 debugfs shows what the fragmentation index for each order is in each zone in
315 the system. Values tending towards 0 imply all 277 the system. Values tending towards 0 imply allocations would fail due to lack
316 of memory, values towards 1000 imply failures 278 of memory, values towards 1000 imply failures are due to fragmentation and -1
317 implies that the allocation will succeed as lo 279 implies that the allocation will succeed as long as watermarks are met.
318 280
319 The kernel will not compact memory in a zone i 281 The kernel will not compact memory in a zone if the
320 fragmentation index is <= extfrag_threshold. T 282 fragmentation index is <= extfrag_threshold. The default value is 500.
321 283
322 284
323 highmem_is_dirtyable 285 highmem_is_dirtyable
324 ==================== 286 ====================
325 287
326 Available only for systems with CONFIG_HIGHMEM 288 Available only for systems with CONFIG_HIGHMEM enabled (32b systems).
327 289
328 This parameter controls whether the high memor 290 This parameter controls whether the high memory is considered for dirty
329 writers throttling. This is not the case by d 291 writers throttling. This is not the case by default which means that
330 only the amount of memory directly visible/usa 292 only the amount of memory directly visible/usable by the kernel can
331 be dirtied. As a result, on systems with a lar 293 be dirtied. As a result, on systems with a large amount of memory and
332 lowmem basically depleted writers might be thr 294 lowmem basically depleted writers might be throttled too early and
333 streaming writes can get very slow. 295 streaming writes can get very slow.
334 296
335 Changing the value to non zero would allow mor 297 Changing the value to non zero would allow more memory to be dirtied
336 and thus allow writers to write more data whic 298 and thus allow writers to write more data which can be flushed to the
337 storage more effectively. Note this also comes 299 storage more effectively. Note this also comes with a risk of pre-mature
338 OOM killer because some writers (e.g. direct b 300 OOM killer because some writers (e.g. direct block device writes) can
339 only use the low memory and they can fill it u 301 only use the low memory and they can fill it up with dirty data without
340 any throttling. 302 any throttling.
341 303
342 304
343 hugetlb_shm_group 305 hugetlb_shm_group
344 ================= 306 =================
345 307
346 hugetlb_shm_group contains group id that is al 308 hugetlb_shm_group contains group id that is allowed to create SysV
347 shared memory segment using hugetlb page. 309 shared memory segment using hugetlb page.
348 310
349 311
350 laptop_mode 312 laptop_mode
351 =========== 313 ===========
352 314
353 laptop_mode is a knob that controls "laptop mo 315 laptop_mode is a knob that controls "laptop mode". All the things that are
354 controlled by this knob are discussed in Docum 316 controlled by this knob are discussed in Documentation/admin-guide/laptops/laptop-mode.rst.
355 317
356 318
357 legacy_va_layout 319 legacy_va_layout
358 ================ 320 ================
359 321
360 If non-zero, this sysctl disables the new 32-b 322 If non-zero, this sysctl disables the new 32-bit mmap layout - the kernel
361 will use the legacy (2.4) layout for all proce 323 will use the legacy (2.4) layout for all processes.
362 324
363 325
364 lowmem_reserve_ratio 326 lowmem_reserve_ratio
365 ==================== 327 ====================
366 328
367 For some specialised workloads on highmem mach 329 For some specialised workloads on highmem machines it is dangerous for
368 the kernel to allow process memory to be alloc 330 the kernel to allow process memory to be allocated from the "lowmem"
369 zone. This is because that memory could then 331 zone. This is because that memory could then be pinned via the mlock()
370 system call, or by unavailability of swapspace 332 system call, or by unavailability of swapspace.
371 333
372 And on large highmem machines this lack of rec 334 And on large highmem machines this lack of reclaimable lowmem memory
373 can be fatal. 335 can be fatal.
374 336
375 So the Linux page allocator has a mechanism wh 337 So the Linux page allocator has a mechanism which prevents allocations
376 which *could* use highmem from using too much 338 which *could* use highmem from using too much lowmem. This means that
377 a certain amount of lowmem is defended from th 339 a certain amount of lowmem is defended from the possibility of being
378 captured into pinned user memory. 340 captured into pinned user memory.
379 341
380 (The same argument applies to the old 16 megab 342 (The same argument applies to the old 16 megabyte ISA DMA region. This
381 mechanism will also defend that region from al 343 mechanism will also defend that region from allocations which could use
382 highmem or lowmem). 344 highmem or lowmem).
383 345
384 The `lowmem_reserve_ratio` tunable determines 346 The `lowmem_reserve_ratio` tunable determines how aggressive the kernel is
385 in defending these lower zones. 347 in defending these lower zones.
386 348
387 If you have a machine which uses highmem or IS 349 If you have a machine which uses highmem or ISA DMA and your
388 applications are using mlock(), or if you are 350 applications are using mlock(), or if you are running with no swap then
389 you probably should change the lowmem_reserve_ 351 you probably should change the lowmem_reserve_ratio setting.
390 352
391 The lowmem_reserve_ratio is an array. You can 353 The lowmem_reserve_ratio is an array. You can see them by reading this file::
392 354
393 % cat /proc/sys/vm/lowmem_reserve_rati 355 % cat /proc/sys/vm/lowmem_reserve_ratio
394 256 256 32 356 256 256 32
395 357
396 But, these values are not used directly. The k 358 But, these values are not used directly. The kernel calculates # of protection
397 pages for each zones from them. These are show 359 pages for each zones from them. These are shown as array of protection pages
398 in /proc/zoneinfo like the following. (This is 360 in /proc/zoneinfo like the following. (This is an example of x86-64 box).
399 Each zone has an array of protection pages lik 361 Each zone has an array of protection pages like this::
400 362
401 Node 0, zone DMA 363 Node 0, zone DMA
402 pages free 1355 364 pages free 1355
403 min 3 365 min 3
404 low 3 366 low 3
405 high 4 367 high 4
406 : 368 :
407 : 369 :
408 numa_other 0 370 numa_other 0
409 protection: (0, 2004, 2004, 2004) 371 protection: (0, 2004, 2004, 2004)
410 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 372 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
411 pagesets 373 pagesets
412 cpu: 0 pcp: 0 374 cpu: 0 pcp: 0
413 : 375 :
414 376
415 These protections are added to score to judge 377 These protections are added to score to judge whether this zone should be used
416 for page allocation or should be reclaimed. 378 for page allocation or should be reclaimed.
417 379
418 In this example, if normal pages (index=2) are 380 In this example, if normal pages (index=2) are required to this DMA zone and
419 watermark[WMARK_HIGH] is used for watermark, t 381 watermark[WMARK_HIGH] is used for watermark, the kernel judges this zone should
420 not be used because pages_free(1355) is smalle 382 not be used because pages_free(1355) is smaller than watermark + protection[2]
421 (4 + 2004 = 2008). If this protection value is 383 (4 + 2004 = 2008). If this protection value is 0, this zone would be used for
422 normal page requirement. If requirement is DMA 384 normal page requirement. If requirement is DMA zone(index=0), protection[0]
423 (=0) is used. 385 (=0) is used.
424 386
425 zone[i]'s protection[j] is calculated by follo 387 zone[i]'s protection[j] is calculated by following expression::
426 388
427 (i < j): 389 (i < j):
428 zone[i]->protection[j] 390 zone[i]->protection[j]
429 = (total sums of managed_pages from zone[i 391 = (total sums of managed_pages from zone[i+1] to zone[j] on the node)
430 / lowmem_reserve_ratio[i]; 392 / lowmem_reserve_ratio[i];
431 (i = j): 393 (i = j):
432 (should not be protected. = 0; 394 (should not be protected. = 0;
433 (i > j): 395 (i > j):
434 (not necessary, but looks 0) 396 (not necessary, but looks 0)
435 397
436 The default values of lowmem_reserve_ratio[i] 398 The default values of lowmem_reserve_ratio[i] are
437 399
438 === ==================================== 400 === ====================================
439 256 (if zone[i] means DMA or DMA32 zone) 401 256 (if zone[i] means DMA or DMA32 zone)
440 32 (others) 402 32 (others)
441 === ==================================== 403 === ====================================
442 404
443 As above expression, they are reciprocal numbe 405 As above expression, they are reciprocal number of ratio.
444 256 means 1/256. # of protection pages becomes 406 256 means 1/256. # of protection pages becomes about "0.39%" of total managed
445 pages of higher zones on the node. 407 pages of higher zones on the node.
446 408
447 If you would like to protect more pages, small 409 If you would like to protect more pages, smaller values are effective.
448 The minimum value is 1 (1/1 -> 100%). The valu 410 The minimum value is 1 (1/1 -> 100%). The value less than 1 completely
449 disables protection of the pages. 411 disables protection of the pages.
450 412
451 413
452 max_map_count: 414 max_map_count:
453 ============== 415 ==============
454 416
455 This file contains the maximum number of memor 417 This file contains the maximum number of memory map areas a process
456 may have. Memory map areas are used as a side- 418 may have. Memory map areas are used as a side-effect of calling
457 malloc, directly by mmap, mprotect, and madvis 419 malloc, directly by mmap, mprotect, and madvise, and also when loading
458 shared libraries. 420 shared libraries.
459 421
460 While most applications need less than a thous 422 While most applications need less than a thousand maps, certain
461 programs, particularly malloc debuggers, may c 423 programs, particularly malloc debuggers, may consume lots of them,
462 e.g., up to one or two maps per allocation. 424 e.g., up to one or two maps per allocation.
463 425
464 The default value is 65530. 426 The default value is 65530.
465 427
466 428
467 mem_profiling 429 mem_profiling
468 ============== 430 ==============
469 431
470 Enable memory profiling (when CONFIG_MEM_ALLOC 432 Enable memory profiling (when CONFIG_MEM_ALLOC_PROFILING=y)
471 433
472 1: Enable memory profiling. 434 1: Enable memory profiling.
473 435
474 0: Disable memory profiling. 436 0: Disable memory profiling.
475 437
476 Enabling memory profiling introduces a small p 438 Enabling memory profiling introduces a small performance overhead for all
477 memory allocations. 439 memory allocations.
478 440
479 The default value depends on CONFIG_MEM_ALLOC_ 441 The default value depends on CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT.
480 442
481 443
482 memory_failure_early_kill: 444 memory_failure_early_kill:
483 ========================== 445 ==========================
484 446
485 Control how to kill processes when uncorrected 447 Control how to kill processes when uncorrected memory error (typically
486 a 2bit error in a memory module) is detected i 448 a 2bit error in a memory module) is detected in the background by hardware
487 that cannot be handled by the kernel. In some 449 that cannot be handled by the kernel. In some cases (like the page
488 still having a valid copy on disk) the kernel 450 still having a valid copy on disk) the kernel will handle the failure
489 transparently without affecting any applicatio 451 transparently without affecting any applications. But if there is
490 no other up-to-date copy of the data it will k 452 no other up-to-date copy of the data it will kill to prevent any data
491 corruptions from propagating. 453 corruptions from propagating.
492 454
493 1: Kill all processes that have the corrupted 455 1: Kill all processes that have the corrupted and not reloadable page mapped
494 as soon as the corruption is detected. Note t 456 as soon as the corruption is detected. Note this is not supported
495 for a few types of pages, like kernel internal 457 for a few types of pages, like kernel internally allocated data or
496 the swap cache, but works for the majority of 458 the swap cache, but works for the majority of user pages.
497 459
498 0: Only unmap the corrupted page from all proc 460 0: Only unmap the corrupted page from all processes and only kill a process
499 who tries to access it. 461 who tries to access it.
500 462
501 The kill is done using a catchable SIGBUS with 463 The kill is done using a catchable SIGBUS with BUS_MCEERR_AO, so processes can
502 handle this if they want to. 464 handle this if they want to.
503 465
504 This is only active on architectures/platforms 466 This is only active on architectures/platforms with advanced machine
505 check handling and depends on the hardware cap 467 check handling and depends on the hardware capabilities.
506 468
507 Applications can override this setting individ 469 Applications can override this setting individually with the PR_MCE_KILL prctl
508 470
509 471
510 memory_failure_recovery 472 memory_failure_recovery
511 ======================= 473 =======================
512 474
513 Enable memory failure recovery (when supported 475 Enable memory failure recovery (when supported by the platform)
514 476
515 1: Attempt recovery. 477 1: Attempt recovery.
516 478
517 0: Always panic on a memory failure. 479 0: Always panic on a memory failure.
518 480
519 481
520 min_free_kbytes 482 min_free_kbytes
521 =============== 483 ===============
522 484
523 This is used to force the Linux VM to keep a m 485 This is used to force the Linux VM to keep a minimum number
524 of kilobytes free. The VM uses this number to 486 of kilobytes free. The VM uses this number to compute a
525 watermark[WMARK_MIN] value for each lowmem zon 487 watermark[WMARK_MIN] value for each lowmem zone in the system.
526 Each lowmem zone gets a number of reserved fre 488 Each lowmem zone gets a number of reserved free pages based
527 proportionally on its size. 489 proportionally on its size.
528 490
529 Some minimal amount of memory is needed to sat 491 Some minimal amount of memory is needed to satisfy PF_MEMALLOC
530 allocations; if you set this to lower than 102 492 allocations; if you set this to lower than 1024KB, your system will
531 become subtly broken, and prone to deadlock un 493 become subtly broken, and prone to deadlock under high loads.
532 494
533 Setting this too high will OOM your machine in 495 Setting this too high will OOM your machine instantly.
534 496
535 497
536 min_slab_ratio 498 min_slab_ratio
537 ============== 499 ==============
538 500
539 This is available only on NUMA kernels. 501 This is available only on NUMA kernels.
540 502
541 A percentage of the total pages in each zone. 503 A percentage of the total pages in each zone. On Zone reclaim
542 (fallback from the local zone occurs) slabs wi 504 (fallback from the local zone occurs) slabs will be reclaimed if more
543 than this percentage of pages in a zone are re 505 than this percentage of pages in a zone are reclaimable slab pages.
544 This insures that the slab growth stays under 506 This insures that the slab growth stays under control even in NUMA
545 systems that rarely perform global reclaim. 507 systems that rarely perform global reclaim.
546 508
547 The default is 5 percent. 509 The default is 5 percent.
548 510
549 Note that slab reclaim is triggered in a per z 511 Note that slab reclaim is triggered in a per zone / node fashion.
550 The process of reclaiming slab memory is curre 512 The process of reclaiming slab memory is currently not node specific
551 and may not be fast. 513 and may not be fast.
552 514
553 515
554 min_unmapped_ratio 516 min_unmapped_ratio
555 ================== 517 ==================
556 518
557 This is available only on NUMA kernels. 519 This is available only on NUMA kernels.
558 520
559 This is a percentage of the total pages in eac 521 This is a percentage of the total pages in each zone. Zone reclaim will
560 only occur if more than this percentage of pag 522 only occur if more than this percentage of pages are in a state that
561 zone_reclaim_mode allows to be reclaimed. 523 zone_reclaim_mode allows to be reclaimed.
562 524
563 If zone_reclaim_mode has the value 4 OR'd, the 525 If zone_reclaim_mode has the value 4 OR'd, then the percentage is compared
564 against all file-backed unmapped pages includi 526 against all file-backed unmapped pages including swapcache pages and tmpfs
565 files. Otherwise, only unmapped pages backed b 527 files. Otherwise, only unmapped pages backed by normal files but not tmpfs
566 files and similar are considered. 528 files and similar are considered.
567 529
568 The default is 1 percent. 530 The default is 1 percent.
569 531
570 532
571 mmap_min_addr 533 mmap_min_addr
572 ============= 534 =============
573 535
574 This file indicates the amount of address spac 536 This file indicates the amount of address space which a user process will
575 be restricted from mmapping. Since kernel nul 537 be restricted from mmapping. Since kernel null dereference bugs could
576 accidentally operate based on the information 538 accidentally operate based on the information in the first couple of pages
577 of memory userspace processes should not be al 539 of memory userspace processes should not be allowed to write to them. By
578 default this value is set to 0 and no protecti 540 default this value is set to 0 and no protections will be enforced by the
579 security module. Setting this value to someth 541 security module. Setting this value to something like 64k will allow the
580 vast majority of applications to work correctl 542 vast majority of applications to work correctly and provide defense in depth
581 against future potential kernel bugs. 543 against future potential kernel bugs.
582 544
583 545
584 mmap_rnd_bits 546 mmap_rnd_bits
585 ============= 547 =============
586 548
587 This value can be used to select the number of 549 This value can be used to select the number of bits to use to
588 determine the random offset to the base addres 550 determine the random offset to the base address of vma regions
589 resulting from mmap allocations on architectur 551 resulting from mmap allocations on architectures which support
590 tuning address space randomization. This valu 552 tuning address space randomization. This value will be bounded
591 by the architecture's minimum and maximum supp 553 by the architecture's minimum and maximum supported values.
592 554
593 This value can be changed after boot using the 555 This value can be changed after boot using the
594 /proc/sys/vm/mmap_rnd_bits tunable 556 /proc/sys/vm/mmap_rnd_bits tunable
595 557
596 558
597 mmap_rnd_compat_bits 559 mmap_rnd_compat_bits
598 ==================== 560 ====================
599 561
600 This value can be used to select the number of 562 This value can be used to select the number of bits to use to
601 determine the random offset to the base addres 563 determine the random offset to the base address of vma regions
602 resulting from mmap allocations for applicatio 564 resulting from mmap allocations for applications run in
603 compatibility mode on architectures which supp 565 compatibility mode on architectures which support tuning address
604 space randomization. This value will be bound 566 space randomization. This value will be bounded by the
605 architecture's minimum and maximum supported v 567 architecture's minimum and maximum supported values.
606 568
607 This value can be changed after boot using the 569 This value can be changed after boot using the
608 /proc/sys/vm/mmap_rnd_compat_bits tunable 570 /proc/sys/vm/mmap_rnd_compat_bits tunable
609 571
610 572
611 nr_hugepages 573 nr_hugepages
612 ============ 574 ============
613 575
614 Change the minimum size of the hugepage pool. 576 Change the minimum size of the hugepage pool.
615 577
616 See Documentation/admin-guide/mm/hugetlbpage.r 578 See Documentation/admin-guide/mm/hugetlbpage.rst
617 579
618 580
619 hugetlb_optimize_vmemmap 581 hugetlb_optimize_vmemmap
620 ======================== 582 ========================
621 583
622 This knob is not available when the size of 's 584 This knob is not available when the size of 'struct page' (a structure defined
623 in include/linux/mm_types.h) is not power of t 585 in include/linux/mm_types.h) is not power of two (an unusual system config could
624 result in this). 586 result in this).
625 587
626 Enable (set to 1) or disable (set to 0) HugeTL 588 Enable (set to 1) or disable (set to 0) HugeTLB Vmemmap Optimization (HVO).
627 589
628 Once enabled, the vmemmap pages of subsequent 590 Once enabled, the vmemmap pages of subsequent allocation of HugeTLB pages from
629 buddy allocator will be optimized (7 pages per 591 buddy allocator will be optimized (7 pages per 2MB HugeTLB page and 4095 pages
630 per 1GB HugeTLB page), whereas already allocat 592 per 1GB HugeTLB page), whereas already allocated HugeTLB pages will not be
631 optimized. When those optimized HugeTLB pages 593 optimized. When those optimized HugeTLB pages are freed from the HugeTLB pool
632 to the buddy allocator, the vmemmap pages repr 594 to the buddy allocator, the vmemmap pages representing that range needs to be
633 remapped again and the vmemmap pages discarded 595 remapped again and the vmemmap pages discarded earlier need to be rellocated
634 again. If your use case is that HugeTLB pages 596 again. If your use case is that HugeTLB pages are allocated 'on the fly' (e.g.
635 never explicitly allocating HugeTLB pages with 597 never explicitly allocating HugeTLB pages with 'nr_hugepages' but only set
636 'nr_overcommit_hugepages', those overcommitted 598 'nr_overcommit_hugepages', those overcommitted HugeTLB pages are allocated 'on
637 the fly') instead of being pulled from the Hug 599 the fly') instead of being pulled from the HugeTLB pool, you should weigh the
638 benefits of memory savings against the more ov 600 benefits of memory savings against the more overhead (~2x slower than before)
639 of allocation or freeing HugeTLB pages between 601 of allocation or freeing HugeTLB pages between the HugeTLB pool and the buddy
640 allocator. Another behavior to note is that i 602 allocator. Another behavior to note is that if the system is under heavy memory
641 pressure, it could prevent the user from freei 603 pressure, it could prevent the user from freeing HugeTLB pages from the HugeTLB
642 pool to the buddy allocator since the allocati 604 pool to the buddy allocator since the allocation of vmemmap pages could be
643 failed, you have to retry later if your system 605 failed, you have to retry later if your system encounter this situation.
644 606
645 Once disabled, the vmemmap pages of subsequent 607 Once disabled, the vmemmap pages of subsequent allocation of HugeTLB pages from
646 buddy allocator will not be optimized meaning 608 buddy allocator will not be optimized meaning the extra overhead at allocation
647 time from buddy allocator disappears, whereas 609 time from buddy allocator disappears, whereas already optimized HugeTLB pages
648 will not be affected. If you want to make sur 610 will not be affected. If you want to make sure there are no optimized HugeTLB
649 pages, you can set "nr_hugepages" to 0 first a 611 pages, you can set "nr_hugepages" to 0 first and then disable this. Note that
650 writing 0 to nr_hugepages will make any "in us 612 writing 0 to nr_hugepages will make any "in use" HugeTLB pages become surplus
651 pages. So, those surplus pages are still opti 613 pages. So, those surplus pages are still optimized until they are no longer
652 in use. You would need to wait for those surp 614 in use. You would need to wait for those surplus pages to be released before
653 there are no optimized pages in the system. 615 there are no optimized pages in the system.
654 616
655 617
656 nr_hugepages_mempolicy 618 nr_hugepages_mempolicy
657 ====================== 619 ======================
658 620
659 Change the size of the hugepage pool at run-ti 621 Change the size of the hugepage pool at run-time on a specific
660 set of NUMA nodes. 622 set of NUMA nodes.
661 623
662 See Documentation/admin-guide/mm/hugetlbpage.r 624 See Documentation/admin-guide/mm/hugetlbpage.rst
663 625
664 626
665 nr_overcommit_hugepages 627 nr_overcommit_hugepages
666 ======================= 628 =======================
667 629
668 Change the maximum size of the hugepage pool. 630 Change the maximum size of the hugepage pool. The maximum is
669 nr_hugepages + nr_overcommit_hugepages. 631 nr_hugepages + nr_overcommit_hugepages.
670 632
671 See Documentation/admin-guide/mm/hugetlbpage.r 633 See Documentation/admin-guide/mm/hugetlbpage.rst
672 634
673 635
674 nr_trim_pages 636 nr_trim_pages
675 ============= 637 =============
676 638
677 This is available only on NOMMU kernels. 639 This is available only on NOMMU kernels.
678 640
679 This value adjusts the excess page trimming be 641 This value adjusts the excess page trimming behaviour of power-of-2 aligned
680 NOMMU mmap allocations. 642 NOMMU mmap allocations.
681 643
682 A value of 0 disables trimming of allocations 644 A value of 0 disables trimming of allocations entirely, while a value of 1
683 trims excess pages aggressively. Any value >= 645 trims excess pages aggressively. Any value >= 1 acts as the watermark where
684 trimming of allocations is initiated. 646 trimming of allocations is initiated.
685 647
686 The default value is 1. 648 The default value is 1.
687 649
688 See Documentation/admin-guide/mm/nommu-mmap.rs 650 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
689 651
690 652
691 numa_zonelist_order 653 numa_zonelist_order
692 =================== 654 ===================
693 655
694 This sysctl is only for NUMA and it is depreca 656 This sysctl is only for NUMA and it is deprecated. Anything but
695 Node order will fail! 657 Node order will fail!
696 658
697 'where the memory is allocated from' is contro 659 'where the memory is allocated from' is controlled by zonelists.
698 660
699 (This documentation ignores ZONE_HIGHMEM/ZONE_ 661 (This documentation ignores ZONE_HIGHMEM/ZONE_DMA32 for simple explanation.
700 you may be able to read ZONE_DMA as ZONE_DMA32 662 you may be able to read ZONE_DMA as ZONE_DMA32...)
701 663
702 In non-NUMA case, a zonelist for GFP_KERNEL is 664 In non-NUMA case, a zonelist for GFP_KERNEL is ordered as following.
703 ZONE_NORMAL -> ZONE_DMA 665 ZONE_NORMAL -> ZONE_DMA
704 This means that a memory allocation request fo 666 This means that a memory allocation request for GFP_KERNEL will
705 get memory from ZONE_DMA only when ZONE_NORMAL 667 get memory from ZONE_DMA only when ZONE_NORMAL is not available.
706 668
707 In NUMA case, you can think of following 2 typ 669 In NUMA case, you can think of following 2 types of order.
708 Assume 2 node NUMA and below is zonelist of No 670 Assume 2 node NUMA and below is zonelist of Node(0)'s GFP_KERNEL::
709 671
710 (A) Node(0) ZONE_NORMAL -> Node(0) ZONE_DMA 672 (A) Node(0) ZONE_NORMAL -> Node(0) ZONE_DMA -> Node(1) ZONE_NORMAL
711 (B) Node(0) ZONE_NORMAL -> Node(1) ZONE_NORM 673 (B) Node(0) ZONE_NORMAL -> Node(1) ZONE_NORMAL -> Node(0) ZONE_DMA.
712 674
713 Type(A) offers the best locality for processes 675 Type(A) offers the best locality for processes on Node(0), but ZONE_DMA
714 will be used before ZONE_NORMAL exhaustion. Th 676 will be used before ZONE_NORMAL exhaustion. This increases possibility of
715 out-of-memory(OOM) of ZONE_DMA because ZONE_DM 677 out-of-memory(OOM) of ZONE_DMA because ZONE_DMA is tend to be small.
716 678
717 Type(B) cannot offer the best locality but is 679 Type(B) cannot offer the best locality but is more robust against OOM of
718 the DMA zone. 680 the DMA zone.
719 681
720 Type(A) is called as "Node" order. Type (B) is 682 Type(A) is called as "Node" order. Type (B) is "Zone" order.
721 683
722 "Node order" orders the zonelists by node, the 684 "Node order" orders the zonelists by node, then by zone within each node.
723 Specify "[Nn]ode" for node order 685 Specify "[Nn]ode" for node order
724 686
725 "Zone Order" orders the zonelists by zone type 687 "Zone Order" orders the zonelists by zone type, then by node within each
726 zone. Specify "[Zz]one" for zone order. 688 zone. Specify "[Zz]one" for zone order.
727 689
728 Specify "[Dd]efault" to request automatic conf 690 Specify "[Dd]efault" to request automatic configuration.
729 691
730 On 32-bit, the Normal zone needs to be preserv 692 On 32-bit, the Normal zone needs to be preserved for allocations accessible
731 by the kernel, so "zone" order will be selecte 693 by the kernel, so "zone" order will be selected.
732 694
733 On 64-bit, devices that require DMA32/DMA are 695 On 64-bit, devices that require DMA32/DMA are relatively rare, so "node"
734 order will be selected. 696 order will be selected.
735 697
736 Default order is recommended unless this is ca 698 Default order is recommended unless this is causing problems for your
737 system/application. 699 system/application.
738 700
739 701
740 oom_dump_tasks 702 oom_dump_tasks
741 ============== 703 ==============
742 704
743 Enables a system-wide task dump (excluding ker 705 Enables a system-wide task dump (excluding kernel threads) to be produced
744 when the kernel performs an OOM-killing and in 706 when the kernel performs an OOM-killing and includes such information as
745 pid, uid, tgid, vm size, rss, pgtables_bytes, 707 pid, uid, tgid, vm size, rss, pgtables_bytes, swapents, oom_score_adj
746 score, and name. This is helpful to determine 708 score, and name. This is helpful to determine why the OOM killer was
747 invoked, to identify the rogue task that cause 709 invoked, to identify the rogue task that caused it, and to determine why
748 the OOM killer chose the task it did to kill. 710 the OOM killer chose the task it did to kill.
749 711
750 If this is set to zero, this information is su 712 If this is set to zero, this information is suppressed. On very
751 large systems with thousands of tasks it may n 713 large systems with thousands of tasks it may not be feasible to dump
752 the memory state information for each one. Su 714 the memory state information for each one. Such systems should not
753 be forced to incur a performance penalty in OO 715 be forced to incur a performance penalty in OOM conditions when the
754 information may not be desired. 716 information may not be desired.
755 717
756 If this is set to non-zero, this information i 718 If this is set to non-zero, this information is shown whenever the
757 OOM killer actually kills a memory-hogging tas 719 OOM killer actually kills a memory-hogging task.
758 720
759 The default value is 1 (enabled). 721 The default value is 1 (enabled).
760 722
761 723
762 oom_kill_allocating_task 724 oom_kill_allocating_task
763 ======================== 725 ========================
764 726
765 This enables or disables killing the OOM-trigg 727 This enables or disables killing the OOM-triggering task in
766 out-of-memory situations. 728 out-of-memory situations.
767 729
768 If this is set to zero, the OOM killer will sc 730 If this is set to zero, the OOM killer will scan through the entire
769 tasklist and select a task based on heuristics 731 tasklist and select a task based on heuristics to kill. This normally
770 selects a rogue memory-hogging task that frees 732 selects a rogue memory-hogging task that frees up a large amount of
771 memory when killed. 733 memory when killed.
772 734
773 If this is set to non-zero, the OOM killer sim 735 If this is set to non-zero, the OOM killer simply kills the task that
774 triggered the out-of-memory condition. This a 736 triggered the out-of-memory condition. This avoids the expensive
775 tasklist scan. 737 tasklist scan.
776 738
777 If panic_on_oom is selected, it takes preceden 739 If panic_on_oom is selected, it takes precedence over whatever value
778 is used in oom_kill_allocating_task. 740 is used in oom_kill_allocating_task.
779 741
780 The default value is 0. 742 The default value is 0.
781 743
782 744
783 overcommit_kbytes 745 overcommit_kbytes
784 ================= 746 =================
785 747
786 When overcommit_memory is set to 2, the commit 748 When overcommit_memory is set to 2, the committed address space is not
787 permitted to exceed swap plus this amount of p 749 permitted to exceed swap plus this amount of physical RAM. See below.
788 750
789 Note: overcommit_kbytes is the counterpart of 751 Note: overcommit_kbytes is the counterpart of overcommit_ratio. Only one
790 of them may be specified at a time. Setting on 752 of them may be specified at a time. Setting one disables the other (which
791 then appears as 0 when read). 753 then appears as 0 when read).
792 754
793 755
794 overcommit_memory 756 overcommit_memory
795 ================= 757 =================
796 758
797 This value contains a flag that enables memory 759 This value contains a flag that enables memory overcommitment.
798 760
799 When this flag is 0, the kernel compares the u 761 When this flag is 0, the kernel compares the userspace memory request
800 size against total memory plus swap and reject 762 size against total memory plus swap and rejects obvious overcommits.
801 763
802 When this flag is 1, the kernel pretends there 764 When this flag is 1, the kernel pretends there is always enough
803 memory until it actually runs out. 765 memory until it actually runs out.
804 766
805 When this flag is 2, the kernel uses a "never 767 When this flag is 2, the kernel uses a "never overcommit"
806 policy that attempts to prevent any overcommit 768 policy that attempts to prevent any overcommit of memory.
807 Note that user_reserve_kbytes affects this pol 769 Note that user_reserve_kbytes affects this policy.
808 770
809 This feature can be very useful because there 771 This feature can be very useful because there are a lot of
810 programs that malloc() huge amounts of memory 772 programs that malloc() huge amounts of memory "just-in-case"
811 and don't use much of it. 773 and don't use much of it.
812 774
813 The default value is 0. 775 The default value is 0.
814 776
815 See Documentation/mm/overcommit-accounting.rst 777 See Documentation/mm/overcommit-accounting.rst and
816 mm/util.c::__vm_enough_memory() for more infor 778 mm/util.c::__vm_enough_memory() for more information.
817 779
818 780
819 overcommit_ratio 781 overcommit_ratio
820 ================ 782 ================
821 783
822 When overcommit_memory is set to 2, the commit 784 When overcommit_memory is set to 2, the committed address
823 space is not permitted to exceed swap plus thi 785 space is not permitted to exceed swap plus this percentage
824 of physical RAM. See above. 786 of physical RAM. See above.
825 787
826 788
827 page-cluster 789 page-cluster
828 ============ 790 ============
829 791
830 page-cluster controls the number of pages up t 792 page-cluster controls the number of pages up to which consecutive pages
831 are read in from swap in a single attempt. Thi 793 are read in from swap in a single attempt. This is the swap counterpart
832 to page cache readahead. 794 to page cache readahead.
833 The mentioned consecutivity is not in terms of 795 The mentioned consecutivity is not in terms of virtual/physical addresses,
834 but consecutive on swap space - that means the 796 but consecutive on swap space - that means they were swapped out together.
835 797
836 It is a logarithmic value - setting it to zero 798 It is a logarithmic value - setting it to zero means "1 page", setting
837 it to 1 means "2 pages", setting it to 2 means 799 it to 1 means "2 pages", setting it to 2 means "4 pages", etc.
838 Zero disables swap readahead completely. 800 Zero disables swap readahead completely.
839 801
840 The default value is three (eight pages at a t 802 The default value is three (eight pages at a time). There may be some
841 small benefits in tuning this to a different v 803 small benefits in tuning this to a different value if your workload is
842 swap-intensive. 804 swap-intensive.
843 805
844 Lower values mean lower latencies for initial 806 Lower values mean lower latencies for initial faults, but at the same time
845 extra faults and I/O delays for following faul 807 extra faults and I/O delays for following faults if they would have been part of
846 that consecutive pages readahead would have br 808 that consecutive pages readahead would have brought in.
847 809
848 810
849 page_lock_unfairness 811 page_lock_unfairness
850 ==================== 812 ====================
851 813
852 This value determines the number of times that 814 This value determines the number of times that the page lock can be
853 stolen from under a waiter. After the lock is 815 stolen from under a waiter. After the lock is stolen the number of times
854 specified in this file (default is 5), the "fa 816 specified in this file (default is 5), the "fair lock handoff" semantics
855 will apply, and the waiter will only be awaken 817 will apply, and the waiter will only be awakened if the lock can be taken.
856 818
857 panic_on_oom 819 panic_on_oom
858 ============ 820 ============
859 821
860 This enables or disables panic on out-of-memor 822 This enables or disables panic on out-of-memory feature.
861 823
862 If this is set to 0, the kernel will kill some 824 If this is set to 0, the kernel will kill some rogue process,
863 called oom_killer. Usually, oom_killer can ki 825 called oom_killer. Usually, oom_killer can kill rogue processes and
864 system will survive. 826 system will survive.
865 827
866 If this is set to 1, the kernel panics when ou 828 If this is set to 1, the kernel panics when out-of-memory happens.
867 However, if a process limits using nodes by me 829 However, if a process limits using nodes by mempolicy/cpusets,
868 and those nodes become memory exhaustion statu 830 and those nodes become memory exhaustion status, one process
869 may be killed by oom-killer. No panic occurs i 831 may be killed by oom-killer. No panic occurs in this case.
870 Because other nodes' memory may be free. This 832 Because other nodes' memory may be free. This means system total status
871 may be not fatal yet. 833 may be not fatal yet.
872 834
873 If this is set to 2, the kernel panics compuls 835 If this is set to 2, the kernel panics compulsorily even on the
874 above-mentioned. Even oom happens under memory 836 above-mentioned. Even oom happens under memory cgroup, the whole
875 system panics. 837 system panics.
876 838
877 The default value is 0. 839 The default value is 0.
878 840
879 1 and 2 are for failover of clustering. Please 841 1 and 2 are for failover of clustering. Please select either
880 according to your policy of failover. 842 according to your policy of failover.
881 843
882 panic_on_oom=2+kdump gives you very strong too 844 panic_on_oom=2+kdump gives you very strong tool to investigate
883 why oom happens. You can get snapshot. 845 why oom happens. You can get snapshot.
884 846
885 847
886 percpu_pagelist_high_fraction 848 percpu_pagelist_high_fraction
887 ============================= 849 =============================
888 850
889 This is the fraction of pages in each zone tha 851 This is the fraction of pages in each zone that are can be stored to
890 per-cpu page lists. It is an upper boundary th 852 per-cpu page lists. It is an upper boundary that is divided depending
891 on the number of online CPUs. The min value fo 853 on the number of online CPUs. The min value for this is 8 which means
892 that we do not allow more than 1/8th of pages 854 that we do not allow more than 1/8th of pages in each zone to be stored
893 on per-cpu page lists. This entry only changes 855 on per-cpu page lists. This entry only changes the value of hot per-cpu
894 page lists. A user can specify a number like 1 856 page lists. A user can specify a number like 100 to allocate 1/100th of
895 each zone between per-cpu lists. 857 each zone between per-cpu lists.
896 858
897 The batch value of each per-cpu page list rema 859 The batch value of each per-cpu page list remains the same regardless of
898 the value of the high fraction so allocation l 860 the value of the high fraction so allocation latencies are unaffected.
899 861
900 The initial value is zero. Kernel uses this va 862 The initial value is zero. Kernel uses this value to set the high pcp->high
901 mark based on the low watermark for the zone a 863 mark based on the low watermark for the zone and the number of local
902 online CPUs. If the user writes '0' to this s 864 online CPUs. If the user writes '0' to this sysctl, it will revert to
903 this default behavior. 865 this default behavior.
904 866
905 867
906 stat_interval 868 stat_interval
907 ============= 869 =============
908 870
909 The time interval between which vm statistics 871 The time interval between which vm statistics are updated. The default
910 is 1 second. 872 is 1 second.
911 873
912 874
913 stat_refresh 875 stat_refresh
914 ============ 876 ============
915 877
916 Any read or write (by root only) flushes all t 878 Any read or write (by root only) flushes all the per-cpu vm statistics
917 into their global totals, for more accurate re 879 into their global totals, for more accurate reports when testing
918 e.g. cat /proc/sys/vm/stat_refresh /proc/memin 880 e.g. cat /proc/sys/vm/stat_refresh /proc/meminfo
919 881
920 As a side-effect, it also checks for negative 882 As a side-effect, it also checks for negative totals (elsewhere reported
921 as 0) and "fails" with EINVAL if any are found 883 as 0) and "fails" with EINVAL if any are found, with a warning in dmesg.
922 (At time of writing, a few stats are known som 884 (At time of writing, a few stats are known sometimes to be found negative,
923 with no ill effects: errors and warnings on th 885 with no ill effects: errors and warnings on these stats are suppressed.)
924 886
925 887
926 numa_stat 888 numa_stat
927 ========= 889 =========
928 890
929 This interface allows runtime configuration of 891 This interface allows runtime configuration of numa statistics.
930 892
931 When page allocation performance becomes a bot 893 When page allocation performance becomes a bottleneck and you can tolerate
932 some possible tool breakage and decreased numa 894 some possible tool breakage and decreased numa counter precision, you can
933 do:: 895 do::
934 896
935 echo 0 > /proc/sys/vm/numa_stat 897 echo 0 > /proc/sys/vm/numa_stat
936 898
937 When page allocation performance is not a bott 899 When page allocation performance is not a bottleneck and you want all
938 tooling to work, you can do:: 900 tooling to work, you can do::
939 901
940 echo 1 > /proc/sys/vm/numa_stat 902 echo 1 > /proc/sys/vm/numa_stat
941 903
942 904
943 swappiness 905 swappiness
944 ========== 906 ==========
945 907
946 This control is used to define the rough relat 908 This control is used to define the rough relative IO cost of swapping
947 and filesystem paging, as a value between 0 an 909 and filesystem paging, as a value between 0 and 200. At 100, the VM
948 assumes equal IO cost and will thus apply memo 910 assumes equal IO cost and will thus apply memory pressure to the page
949 cache and swap-backed pages equally; lower val 911 cache and swap-backed pages equally; lower values signify more
950 expensive swap IO, higher values indicates che 912 expensive swap IO, higher values indicates cheaper.
951 913
952 Keep in mind that filesystem IO patterns under 914 Keep in mind that filesystem IO patterns under memory pressure tend to
953 be more efficient than swap's random IO. An op 915 be more efficient than swap's random IO. An optimal value will require
954 experimentation and will also be workload-depe 916 experimentation and will also be workload-dependent.
955 917
956 The default value is 60. 918 The default value is 60.
957 919
958 For in-memory swap, like zram or zswap, as wel 920 For in-memory swap, like zram or zswap, as well as hybrid setups that
959 have swap on faster devices than the filesyste 921 have swap on faster devices than the filesystem, values beyond 100 can
960 be considered. For example, if the random IO a 922 be considered. For example, if the random IO against the swap device
961 is on average 2x faster than IO from the files 923 is on average 2x faster than IO from the filesystem, swappiness should
962 be 133 (x + 2x = 200, 2x = 133.33). 924 be 133 (x + 2x = 200, 2x = 133.33).
963 925
964 At 0, the kernel will not initiate swap until 926 At 0, the kernel will not initiate swap until the amount of free and
965 file-backed pages is less than the high waterm 927 file-backed pages is less than the high watermark in a zone.
966 928
967 929
968 unprivileged_userfaultfd 930 unprivileged_userfaultfd
969 ======================== 931 ========================
970 932
971 This flag controls the mode in which unprivile 933 This flag controls the mode in which unprivileged users can use the
972 userfaultfd system calls. Set this to 0 to res 934 userfaultfd system calls. Set this to 0 to restrict unprivileged users
973 to handle page faults in user mode only. In th 935 to handle page faults in user mode only. In this case, users without
974 SYS_CAP_PTRACE must pass UFFD_USER_MODE_ONLY i 936 SYS_CAP_PTRACE must pass UFFD_USER_MODE_ONLY in order for userfaultfd to
975 succeed. Prohibiting use of userfaultfd for ha 937 succeed. Prohibiting use of userfaultfd for handling faults from kernel
976 mode may make certain vulnerabilities more dif 938 mode may make certain vulnerabilities more difficult to exploit.
977 939
978 Set this to 1 to allow unprivileged users to u 940 Set this to 1 to allow unprivileged users to use the userfaultfd system
979 calls without any restrictions. 941 calls without any restrictions.
980 942
981 The default value is 0. 943 The default value is 0.
982 944
983 Another way to control permissions for userfau 945 Another way to control permissions for userfaultfd is to use
984 /dev/userfaultfd instead of userfaultfd(2). Se 946 /dev/userfaultfd instead of userfaultfd(2). See
985 Documentation/admin-guide/mm/userfaultfd.rst. 947 Documentation/admin-guide/mm/userfaultfd.rst.
986 948
987 user_reserve_kbytes 949 user_reserve_kbytes
988 =================== 950 ===================
989 951
990 When overcommit_memory is set to 2, "never ove 952 When overcommit_memory is set to 2, "never overcommit" mode, reserve
991 min(3% of current process size, user_reserve_k 953 min(3% of current process size, user_reserve_kbytes) of free memory.
992 This is intended to prevent a user from starti 954 This is intended to prevent a user from starting a single memory hogging
993 process, such that they cannot recover (kill t 955 process, such that they cannot recover (kill the hog).
994 956
995 user_reserve_kbytes defaults to min(3% of the 957 user_reserve_kbytes defaults to min(3% of the current process size, 128MB).
996 958
997 If this is reduced to zero, then the user will 959 If this is reduced to zero, then the user will be allowed to allocate
998 all free memory with a single process, minus a 960 all free memory with a single process, minus admin_reserve_kbytes.
999 Any subsequent attempts to execute a command w 961 Any subsequent attempts to execute a command will result in
1000 "fork: Cannot allocate memory". 962 "fork: Cannot allocate memory".
1001 963
1002 Changing this takes effect whenever an applic 964 Changing this takes effect whenever an application requests memory.
1003 965
1004 966
1005 vfs_cache_pressure 967 vfs_cache_pressure
1006 ================== 968 ==================
1007 969
1008 This percentage value controls the tendency o 970 This percentage value controls the tendency of the kernel to reclaim
1009 the memory which is used for caching of direc 971 the memory which is used for caching of directory and inode objects.
1010 972
1011 At the default value of vfs_cache_pressure=10 973 At the default value of vfs_cache_pressure=100 the kernel will attempt to
1012 reclaim dentries and inodes at a "fair" rate 974 reclaim dentries and inodes at a "fair" rate with respect to pagecache and
1013 swapcache reclaim. Decreasing vfs_cache_pres 975 swapcache reclaim. Decreasing vfs_cache_pressure causes the kernel to prefer
1014 to retain dentry and inode caches. When vfs_c 976 to retain dentry and inode caches. When vfs_cache_pressure=0, the kernel will
1015 never reclaim dentries and inodes due to memo 977 never reclaim dentries and inodes due to memory pressure and this can easily
1016 lead to out-of-memory conditions. Increasing 978 lead to out-of-memory conditions. Increasing vfs_cache_pressure beyond 100
1017 causes the kernel to prefer to reclaim dentri 979 causes the kernel to prefer to reclaim dentries and inodes.
1018 980
1019 Increasing vfs_cache_pressure significantly b 981 Increasing vfs_cache_pressure significantly beyond 100 may have negative
1020 performance impact. Reclaim code needs to tak 982 performance impact. Reclaim code needs to take various locks to find freeable
1021 directory and inode objects. With vfs_cache_p 983 directory and inode objects. With vfs_cache_pressure=1000, it will look for
1022 ten times more freeable objects than there ar 984 ten times more freeable objects than there are.
1023 985
1024 986
1025 watermark_boost_factor 987 watermark_boost_factor
1026 ====================== 988 ======================
1027 989
1028 This factor controls the level of reclaim whe 990 This factor controls the level of reclaim when memory is being fragmented.
1029 It defines the percentage of the high waterma 991 It defines the percentage of the high watermark of a zone that will be
1030 reclaimed if pages of different mobility are 992 reclaimed if pages of different mobility are being mixed within pageblocks.
1031 The intent is that compaction has less work t 993 The intent is that compaction has less work to do in the future and to
1032 increase the success rate of future high-orde 994 increase the success rate of future high-order allocations such as SLUB
1033 allocations, THP and hugetlbfs pages. 995 allocations, THP and hugetlbfs pages.
1034 996
1035 To make it sensible with respect to the water 997 To make it sensible with respect to the watermark_scale_factor
1036 parameter, the unit is in fractions of 10,000 998 parameter, the unit is in fractions of 10,000. The default value of
1037 15,000 means that up to 150% of the high wate 999 15,000 means that up to 150% of the high watermark will be reclaimed in the
1038 event of a pageblock being mixed due to fragm 1000 event of a pageblock being mixed due to fragmentation. The level of reclaim
1039 is determined by the number of fragmentation 1001 is determined by the number of fragmentation events that occurred in the
1040 recent past. If this value is smaller than a 1002 recent past. If this value is smaller than a pageblock then a pageblocks
1041 worth of pages will be reclaimed (e.g. 2MB o 1003 worth of pages will be reclaimed (e.g. 2MB on 64-bit x86). A boost factor
1042 of 0 will disable the feature. 1004 of 0 will disable the feature.
1043 1005
1044 1006
1045 watermark_scale_factor 1007 watermark_scale_factor
1046 ====================== 1008 ======================
1047 1009
1048 This factor controls the aggressiveness of ks 1010 This factor controls the aggressiveness of kswapd. It defines the
1049 amount of memory left in a node/system before 1011 amount of memory left in a node/system before kswapd is woken up and
1050 how much memory needs to be free before kswap 1012 how much memory needs to be free before kswapd goes back to sleep.
1051 1013
1052 The unit is in fractions of 10,000. The defau 1014 The unit is in fractions of 10,000. The default value of 10 means the
1053 distances between watermarks are 0.1% of the 1015 distances between watermarks are 0.1% of the available memory in the
1054 node/system. The maximum value is 3000, or 30 1016 node/system. The maximum value is 3000, or 30% of memory.
1055 1017
1056 A high rate of threads entering direct reclai 1018 A high rate of threads entering direct reclaim (allocstall) or kswapd
1057 going to sleep prematurely (kswapd_low_wmark_ 1019 going to sleep prematurely (kswapd_low_wmark_hit_quickly) can indicate
1058 that the number of free pages kswapd maintain 1020 that the number of free pages kswapd maintains for latency reasons is
1059 too small for the allocation bursts occurring 1021 too small for the allocation bursts occurring in the system. This knob
1060 can then be used to tune kswapd aggressivenes 1022 can then be used to tune kswapd aggressiveness accordingly.
1061 1023
1062 1024
1063 zone_reclaim_mode 1025 zone_reclaim_mode
1064 ================= 1026 =================
1065 1027
1066 Zone_reclaim_mode allows someone to set more 1028 Zone_reclaim_mode allows someone to set more or less aggressive approaches to
1067 reclaim memory when a zone runs out of memory 1029 reclaim memory when a zone runs out of memory. If it is set to zero then no
1068 zone reclaim occurs. Allocations will be sati 1030 zone reclaim occurs. Allocations will be satisfied from other zones / nodes
1069 in the system. 1031 in the system.
1070 1032
1071 This is value OR'ed together of 1033 This is value OR'ed together of
1072 1034
1073 = =================================== 1035 = ===================================
1074 1 Zone reclaim on 1036 1 Zone reclaim on
1075 2 Zone reclaim writes dirty pages out 1037 2 Zone reclaim writes dirty pages out
1076 4 Zone reclaim swaps pages 1038 4 Zone reclaim swaps pages
1077 = =================================== 1039 = ===================================
1078 1040
1079 zone_reclaim_mode is disabled by default. Fo 1041 zone_reclaim_mode is disabled by default. For file servers or workloads
1080 that benefit from having their data cached, z 1042 that benefit from having their data cached, zone_reclaim_mode should be
1081 left disabled as the caching effect is likely 1043 left disabled as the caching effect is likely to be more important than
1082 data locality. 1044 data locality.
1083 1045
1084 Consider enabling one or more zone_reclaim mo 1046 Consider enabling one or more zone_reclaim mode bits if it's known that the
1085 workload is partitioned such that each partit 1047 workload is partitioned such that each partition fits within a NUMA node
1086 and that accessing remote memory would cause 1048 and that accessing remote memory would cause a measurable performance
1087 reduction. The page allocator will take addi 1049 reduction. The page allocator will take additional actions before
1088 allocating off node pages. 1050 allocating off node pages.
1089 1051
1090 Allowing zone reclaim to write out pages stop 1052 Allowing zone reclaim to write out pages stops processes that are
1091 writing large amounts of data from dirtying p 1053 writing large amounts of data from dirtying pages on other nodes. Zone
1092 reclaim will write out dirty pages if a zone 1054 reclaim will write out dirty pages if a zone fills up and so effectively
1093 throttle the process. This may decrease the p 1055 throttle the process. This may decrease the performance of a single process
1094 since it cannot use all of system memory to b 1056 since it cannot use all of system memory to buffer the outgoing writes
1095 anymore but it preserve the memory on other n 1057 anymore but it preserve the memory on other nodes so that the performance
1096 of other processes running on other nodes wil 1058 of other processes running on other nodes will not be affected.
1097 1059
1098 Allowing regular swap effectively restricts a 1060 Allowing regular swap effectively restricts allocations to the local
1099 node unless explicitly overridden by memory p 1061 node unless explicitly overridden by memory policies or cpuset
1100 configurations. 1062 configurations.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.