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