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 - 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 <<
42 - hugetlb_shm_group 40 - hugetlb_shm_group
43 - laptop_mode 41 - laptop_mode
44 - legacy_va_layout 42 - legacy_va_layout
45 - lowmem_reserve_ratio 43 - lowmem_reserve_ratio
46 - max_map_count 44 - max_map_count
47 - mem_profiling (only if CONFIG_MEM_AL <<
48 - memory_failure_early_kill 45 - memory_failure_early_kill
49 - memory_failure_recovery 46 - memory_failure_recovery
50 - min_free_kbytes 47 - min_free_kbytes
51 - min_slab_ratio 48 - min_slab_ratio
52 - min_unmapped_ratio 49 - min_unmapped_ratio
53 - mmap_min_addr 50 - mmap_min_addr
54 - mmap_rnd_bits 51 - mmap_rnd_bits
55 - mmap_rnd_compat_bits 52 - mmap_rnd_compat_bits
56 - nr_hugepages 53 - nr_hugepages
57 - nr_hugepages_mempolicy 54 - nr_hugepages_mempolicy
58 - nr_overcommit_hugepages 55 - nr_overcommit_hugepages
59 - nr_trim_pages (only if CONFIG_MMU=n) 56 - nr_trim_pages (only if CONFIG_MMU=n)
60 - numa_zonelist_order 57 - numa_zonelist_order
61 - oom_dump_tasks 58 - oom_dump_tasks
62 - oom_kill_allocating_task 59 - oom_kill_allocating_task
63 - overcommit_kbytes 60 - overcommit_kbytes
64 - overcommit_memory 61 - overcommit_memory
65 - overcommit_ratio 62 - overcommit_ratio
66 - page-cluster 63 - page-cluster
67 - page_lock_unfairness <<
68 - panic_on_oom 64 - panic_on_oom
69 - percpu_pagelist_high_fraction !! 65 - percpu_pagelist_fraction
70 - stat_interval 66 - stat_interval
71 - stat_refresh 67 - stat_refresh
72 - numa_stat 68 - numa_stat
73 - swappiness 69 - swappiness
74 - unprivileged_userfaultfd 70 - unprivileged_userfaultfd
75 - user_reserve_kbytes 71 - user_reserve_kbytes
76 - vfs_cache_pressure 72 - vfs_cache_pressure
77 - watermark_boost_factor 73 - watermark_boost_factor
78 - watermark_scale_factor 74 - watermark_scale_factor
79 - zone_reclaim_mode 75 - zone_reclaim_mode
80 76
81 77
82 admin_reserve_kbytes 78 admin_reserve_kbytes
83 ==================== 79 ====================
84 80
85 The amount of free memory in the system that s 81 The amount of free memory in the system that should be reserved for users
86 with the capability cap_sys_admin. 82 with the capability cap_sys_admin.
87 83
88 admin_reserve_kbytes defaults to min(3% of fre 84 admin_reserve_kbytes defaults to min(3% of free pages, 8MB)
89 85
90 That should provide enough for the admin to lo 86 That should provide enough for the admin to log in and kill a process,
91 if necessary, under the default overcommit 'gu 87 if necessary, under the default overcommit 'guess' mode.
92 88
93 Systems running under overcommit 'never' shoul 89 Systems running under overcommit 'never' should increase this to account
94 for the full Virtual Memory Size of programs u 90 for the full Virtual Memory Size of programs used to recover. Otherwise,
95 root may not be able to log in to recover the 91 root may not be able to log in to recover the system.
96 92
97 How do you calculate a minimum useful reserve? 93 How do you calculate a minimum useful reserve?
98 94
99 sshd or login + bash (or some other shell) + t 95 sshd or login + bash (or some other shell) + top (or ps, kill, etc.)
100 96
101 For overcommit 'guess', we can sum resident se 97 For overcommit 'guess', we can sum resident set sizes (RSS).
102 On x86_64 this is about 8MB. 98 On x86_64 this is about 8MB.
103 99
104 For overcommit 'never', we can take the max of 100 For overcommit 'never', we can take the max of their virtual sizes (VSZ)
105 and add the sum of their RSS. 101 and add the sum of their RSS.
106 On x86_64 this is about 128MB. 102 On x86_64 this is about 128MB.
107 103
108 Changing this takes effect whenever an applica 104 Changing this takes effect whenever an application requests memory.
109 105
110 106
>> 107 block_dump
>> 108 ==========
>> 109
>> 110 block_dump enables block I/O debugging when set to a nonzero value. More
>> 111 information on block I/O debugging is in Documentation/admin-guide/laptops/laptop-mode.rst.
>> 112
>> 113
111 compact_memory 114 compact_memory
112 ============== 115 ==============
113 116
114 Available only when CONFIG_COMPACTION is set. 117 Available only when CONFIG_COMPACTION is set. When 1 is written to the file,
115 all zones are compacted such that free memory 118 all zones are compacted such that free memory is available in contiguous
116 blocks where possible. This can be important f 119 blocks where possible. This can be important for example in the allocation of
117 huge pages although processes will also direct 120 huge pages although processes will also directly compact memory as required.
118 121
119 compaction_proactiveness 122 compaction_proactiveness
120 ======================== 123 ========================
121 124
122 This tunable takes a value in the range [0, 10 125 This tunable takes a value in the range [0, 100] with a default value of
123 20. This tunable determines how aggressively c 126 20. This tunable determines how aggressively compaction is done in the
124 background. Write of a non zero value to this !! 127 background. Setting it to 0 disables proactive compaction.
125 trigger the proactive compaction. Setting it t <<
126 128
127 Note that compaction has a non-trivial system- 129 Note that compaction has a non-trivial system-wide impact as pages
128 belonging to different processes are moved aro 130 belonging to different processes are moved around, which could also lead
129 to latency spikes in unsuspecting applications 131 to latency spikes in unsuspecting applications. The kernel employs
130 various heuristics to avoid wasting CPU cycles 132 various heuristics to avoid wasting CPU cycles if it detects that
131 proactive compaction is not being effective. 133 proactive compaction is not being effective.
132 134
133 Be careful when setting it to extreme values l 135 Be careful when setting it to extreme values like 100, as that may
134 cause excessive background compaction activity 136 cause excessive background compaction activity.
135 137
136 compact_unevictable_allowed 138 compact_unevictable_allowed
137 =========================== 139 ===========================
138 140
139 Available only when CONFIG_COMPACTION is set. 141 Available only when CONFIG_COMPACTION is set. When set to 1, compaction is
140 allowed to examine the unevictable lru (mlocke 142 allowed to examine the unevictable lru (mlocked pages) for pages to compact.
141 This should be used on systems where stalls fo 143 This should be used on systems where stalls for minor page faults are an
142 acceptable trade for large contiguous free mem 144 acceptable trade for large contiguous free memory. Set to 0 to prevent
143 compaction from moving pages that are unevicta 145 compaction from moving pages that are unevictable. Default value is 1.
144 On CONFIG_PREEMPT_RT the default value is 0 in 146 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 !! 147 to compaction, which would block the task from becomming active until the fault
146 is resolved. 148 is resolved.
147 149
148 150
149 dirty_background_bytes 151 dirty_background_bytes
150 ====================== 152 ======================
151 153
152 Contains the amount of dirty memory at which t 154 Contains the amount of dirty memory at which the background kernel
153 flusher threads will start writeback. 155 flusher threads will start writeback.
154 156
155 Note: 157 Note:
156 dirty_background_bytes is the counterpart of 158 dirty_background_bytes is the counterpart of dirty_background_ratio. Only
157 one of them may be specified at a time. When 159 one of them may be specified at a time. When one sysctl is written it is
158 immediately taken into account to evaluate t 160 immediately taken into account to evaluate the dirty memory limits and the
159 other appears as 0 when read. 161 other appears as 0 when read.
160 162
161 163
162 dirty_background_ratio 164 dirty_background_ratio
163 ====================== 165 ======================
164 166
165 Contains, as a percentage of total available m 167 Contains, as a percentage of total available memory that contains free pages
166 and reclaimable pages, the number of pages at 168 and reclaimable pages, the number of pages at which the background kernel
167 flusher threads will start writing out dirty d 169 flusher threads will start writing out dirty data.
168 170
169 The total available memory is not equal to tot 171 The total available memory is not equal to total system memory.
170 172
171 173
172 dirty_bytes 174 dirty_bytes
173 =========== 175 ===========
174 176
175 Contains the amount of dirty memory at which a 177 Contains the amount of dirty memory at which a process generating disk writes
176 will itself start writeback. 178 will itself start writeback.
177 179
178 Note: dirty_bytes is the counterpart of dirty_ 180 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 181 specified at a time. When one sysctl is written it is immediately taken into
180 account to evaluate the dirty memory limits an 182 account to evaluate the dirty memory limits and the other appears as 0 when
181 read. 183 read.
182 184
183 Note: the minimum value allowed for dirty_byte 185 Note: the minimum value allowed for dirty_bytes is two pages (in bytes); any
184 value lower than this limit will be ignored an 186 value lower than this limit will be ignored and the old configuration will be
185 retained. 187 retained.
186 188
187 189
188 dirty_expire_centisecs 190 dirty_expire_centisecs
189 ====================== 191 ======================
190 192
191 This tunable is used to define when dirty data 193 This tunable is used to define when dirty data is old enough to be eligible
192 for writeout by the kernel flusher threads. I 194 for writeout by the kernel flusher threads. It is expressed in 100'ths
193 of a second. Data which has been dirty in-mem 195 of a second. Data which has been dirty in-memory for longer than this
194 interval will be written out next time a flush 196 interval will be written out next time a flusher thread wakes up.
195 197
196 198
197 dirty_ratio 199 dirty_ratio
198 =========== 200 ===========
199 201
200 Contains, as a percentage of total available m 202 Contains, as a percentage of total available memory that contains free pages
201 and reclaimable pages, the number of pages at 203 and reclaimable pages, the number of pages at which a process which is
202 generating disk writes will itself start writi 204 generating disk writes will itself start writing out dirty data.
203 205
204 The total available memory is not equal to tot 206 The total available memory is not equal to total system memory.
205 207
206 208
207 dirtytime_expire_seconds 209 dirtytime_expire_seconds
208 ======================== 210 ========================
209 211
210 When a lazytime inode is constantly having its 212 When a lazytime inode is constantly having its pages dirtied, the inode with
211 an updated timestamp will never get chance to 213 an updated timestamp will never get chance to be written out. And, if the
212 only thing that has happened on the file syste 214 only thing that has happened on the file system is a dirtytime inode caused
213 by an atime update, a worker will be scheduled 215 by an atime update, a worker will be scheduled to make sure that inode
214 eventually gets pushed out to disk. This tuna 216 eventually gets pushed out to disk. This tunable is used to define when dirty
215 inode is old enough to be eligible for writeba 217 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 218 And, it is also used as the interval to wakeup dirtytime_writeback thread.
217 219
218 220
219 dirty_writeback_centisecs 221 dirty_writeback_centisecs
220 ========================= 222 =========================
221 223
222 The kernel flusher threads will periodically w 224 The kernel flusher threads will periodically wake up and write `old` data
223 out to disk. This tunable expresses the inter 225 out to disk. This tunable expresses the interval between those wakeups, in
224 100'ths of a second. 226 100'ths of a second.
225 227
226 Setting this to zero disables periodic writeba 228 Setting this to zero disables periodic writeback altogether.
227 229
228 230
229 drop_caches 231 drop_caches
230 =========== 232 ===========
231 233
232 Writing to this will cause the kernel to drop 234 Writing to this will cause the kernel to drop clean caches, as well as
233 reclaimable slab objects like dentries and ino 235 reclaimable slab objects like dentries and inodes. Once dropped, their
234 memory becomes free. 236 memory becomes free.
235 237
236 To free pagecache:: 238 To free pagecache::
237 239
238 echo 1 > /proc/sys/vm/drop_caches 240 echo 1 > /proc/sys/vm/drop_caches
239 241
240 To free reclaimable slab objects (includes den 242 To free reclaimable slab objects (includes dentries and inodes)::
241 243
242 echo 2 > /proc/sys/vm/drop_caches 244 echo 2 > /proc/sys/vm/drop_caches
243 245
244 To free slab objects and pagecache:: 246 To free slab objects and pagecache::
245 247
246 echo 3 > /proc/sys/vm/drop_caches 248 echo 3 > /proc/sys/vm/drop_caches
247 249
248 This is a non-destructive operation and will n 250 This is a non-destructive operation and will not free any dirty objects.
249 To increase the number of objects freed by thi 251 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 252 `sync` prior to writing to /proc/sys/vm/drop_caches. This will minimize the
251 number of dirty objects on the system and crea 253 number of dirty objects on the system and create more candidates to be
252 dropped. 254 dropped.
253 255
254 This file is not a means to control the growth 256 This file is not a means to control the growth of the various kernel caches
255 (inodes, dentries, pagecache, etc...) These o 257 (inodes, dentries, pagecache, etc...) These objects are automatically
256 reclaimed by the kernel when memory is needed 258 reclaimed by the kernel when memory is needed elsewhere on the system.
257 259
258 Use of this file can cause performance problem 260 Use of this file can cause performance problems. Since it discards cached
259 objects, it may cost a significant amount of I 261 objects, it may cost a significant amount of I/O and CPU to recreate the
260 dropped objects, especially if they were under 262 dropped objects, especially if they were under heavy use. Because of this,
261 use outside of a testing or debugging environm 263 use outside of a testing or debugging environment is not recommended.
262 264
263 You may see informational messages in your ker 265 You may see informational messages in your kernel log when this file is
264 used:: 266 used::
265 267
266 cat (1234): drop_caches: 3 268 cat (1234): drop_caches: 3
267 269
268 These are informational only. They do not mea 270 These are informational only. They do not mean that anything is wrong
269 with your system. To disable them, echo 4 (bi 271 with your system. To disable them, echo 4 (bit 2) into drop_caches.
270 272
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 <<
308 273
309 extfrag_threshold 274 extfrag_threshold
310 ================= 275 =================
311 276
312 This parameter affects whether the kernel will 277 This parameter affects whether the kernel will compact memory or direct
313 reclaim to satisfy a high-order allocation. Th 278 reclaim to satisfy a high-order allocation. The extfrag/extfrag_index file in
314 debugfs shows what the fragmentation index for 279 debugfs shows what the fragmentation index for each order is in each zone in
315 the system. Values tending towards 0 imply all 280 the system. Values tending towards 0 imply allocations would fail due to lack
316 of memory, values towards 1000 imply failures 281 of memory, values towards 1000 imply failures are due to fragmentation and -1
317 implies that the allocation will succeed as lo 282 implies that the allocation will succeed as long as watermarks are met.
318 283
319 The kernel will not compact memory in a zone i 284 The kernel will not compact memory in a zone if the
320 fragmentation index is <= extfrag_threshold. T 285 fragmentation index is <= extfrag_threshold. The default value is 500.
321 286
322 287
323 highmem_is_dirtyable 288 highmem_is_dirtyable
324 ==================== 289 ====================
325 290
326 Available only for systems with CONFIG_HIGHMEM 291 Available only for systems with CONFIG_HIGHMEM enabled (32b systems).
327 292
328 This parameter controls whether the high memor 293 This parameter controls whether the high memory is considered for dirty
329 writers throttling. This is not the case by d 294 writers throttling. This is not the case by default which means that
330 only the amount of memory directly visible/usa 295 only the amount of memory directly visible/usable by the kernel can
331 be dirtied. As a result, on systems with a lar 296 be dirtied. As a result, on systems with a large amount of memory and
332 lowmem basically depleted writers might be thr 297 lowmem basically depleted writers might be throttled too early and
333 streaming writes can get very slow. 298 streaming writes can get very slow.
334 299
335 Changing the value to non zero would allow mor 300 Changing the value to non zero would allow more memory to be dirtied
336 and thus allow writers to write more data whic 301 and thus allow writers to write more data which can be flushed to the
337 storage more effectively. Note this also comes 302 storage more effectively. Note this also comes with a risk of pre-mature
338 OOM killer because some writers (e.g. direct b 303 OOM killer because some writers (e.g. direct block device writes) can
339 only use the low memory and they can fill it u 304 only use the low memory and they can fill it up with dirty data without
340 any throttling. 305 any throttling.
341 306
342 307
343 hugetlb_shm_group 308 hugetlb_shm_group
344 ================= 309 =================
345 310
346 hugetlb_shm_group contains group id that is al 311 hugetlb_shm_group contains group id that is allowed to create SysV
347 shared memory segment using hugetlb page. 312 shared memory segment using hugetlb page.
348 313
349 314
350 laptop_mode 315 laptop_mode
351 =========== 316 ===========
352 317
353 laptop_mode is a knob that controls "laptop mo 318 laptop_mode is a knob that controls "laptop mode". All the things that are
354 controlled by this knob are discussed in Docum 319 controlled by this knob are discussed in Documentation/admin-guide/laptops/laptop-mode.rst.
355 320
356 321
357 legacy_va_layout 322 legacy_va_layout
358 ================ 323 ================
359 324
360 If non-zero, this sysctl disables the new 32-b 325 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 326 will use the legacy (2.4) layout for all processes.
362 327
363 328
364 lowmem_reserve_ratio 329 lowmem_reserve_ratio
365 ==================== 330 ====================
366 331
367 For some specialised workloads on highmem mach 332 For some specialised workloads on highmem machines it is dangerous for
368 the kernel to allow process memory to be alloc 333 the kernel to allow process memory to be allocated from the "lowmem"
369 zone. This is because that memory could then 334 zone. This is because that memory could then be pinned via the mlock()
370 system call, or by unavailability of swapspace 335 system call, or by unavailability of swapspace.
371 336
372 And on large highmem machines this lack of rec 337 And on large highmem machines this lack of reclaimable lowmem memory
373 can be fatal. 338 can be fatal.
374 339
375 So the Linux page allocator has a mechanism wh 340 So the Linux page allocator has a mechanism which prevents allocations
376 which *could* use highmem from using too much 341 which *could* use highmem from using too much lowmem. This means that
377 a certain amount of lowmem is defended from th 342 a certain amount of lowmem is defended from the possibility of being
378 captured into pinned user memory. 343 captured into pinned user memory.
379 344
380 (The same argument applies to the old 16 megab 345 (The same argument applies to the old 16 megabyte ISA DMA region. This
381 mechanism will also defend that region from al 346 mechanism will also defend that region from allocations which could use
382 highmem or lowmem). 347 highmem or lowmem).
383 348
384 The `lowmem_reserve_ratio` tunable determines 349 The `lowmem_reserve_ratio` tunable determines how aggressive the kernel is
385 in defending these lower zones. 350 in defending these lower zones.
386 351
387 If you have a machine which uses highmem or IS 352 If you have a machine which uses highmem or ISA DMA and your
388 applications are using mlock(), or if you are 353 applications are using mlock(), or if you are running with no swap then
389 you probably should change the lowmem_reserve_ 354 you probably should change the lowmem_reserve_ratio setting.
390 355
391 The lowmem_reserve_ratio is an array. You can 356 The lowmem_reserve_ratio is an array. You can see them by reading this file::
392 357
393 % cat /proc/sys/vm/lowmem_reserve_rati 358 % cat /proc/sys/vm/lowmem_reserve_ratio
394 256 256 32 359 256 256 32
395 360
396 But, these values are not used directly. The k 361 But, these values are not used directly. The kernel calculates # of protection
397 pages for each zones from them. These are show 362 pages for each zones from them. These are shown as array of protection pages
398 in /proc/zoneinfo like the following. (This is !! 363 in /proc/zoneinfo like followings. (This is an example of x86-64 box).
399 Each zone has an array of protection pages lik 364 Each zone has an array of protection pages like this::
400 365
401 Node 0, zone DMA 366 Node 0, zone DMA
402 pages free 1355 367 pages free 1355
403 min 3 368 min 3
404 low 3 369 low 3
405 high 4 370 high 4
406 : 371 :
407 : 372 :
408 numa_other 0 373 numa_other 0
409 protection: (0, 2004, 2004, 2004) 374 protection: (0, 2004, 2004, 2004)
410 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 375 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
411 pagesets 376 pagesets
412 cpu: 0 pcp: 0 377 cpu: 0 pcp: 0
413 : 378 :
414 379
415 These protections are added to score to judge 380 These protections are added to score to judge whether this zone should be used
416 for page allocation or should be reclaimed. 381 for page allocation or should be reclaimed.
417 382
418 In this example, if normal pages (index=2) are 383 In this example, if normal pages (index=2) are required to this DMA zone and
419 watermark[WMARK_HIGH] is used for watermark, t 384 watermark[WMARK_HIGH] is used for watermark, the kernel judges this zone should
420 not be used because pages_free(1355) is smalle 385 not be used because pages_free(1355) is smaller than watermark + protection[2]
421 (4 + 2004 = 2008). If this protection value is 386 (4 + 2004 = 2008). If this protection value is 0, this zone would be used for
422 normal page requirement. If requirement is DMA 387 normal page requirement. If requirement is DMA zone(index=0), protection[0]
423 (=0) is used. 388 (=0) is used.
424 389
425 zone[i]'s protection[j] is calculated by follo 390 zone[i]'s protection[j] is calculated by following expression::
426 391
427 (i < j): 392 (i < j):
428 zone[i]->protection[j] 393 zone[i]->protection[j]
429 = (total sums of managed_pages from zone[i 394 = (total sums of managed_pages from zone[i+1] to zone[j] on the node)
430 / lowmem_reserve_ratio[i]; 395 / lowmem_reserve_ratio[i];
431 (i = j): 396 (i = j):
432 (should not be protected. = 0; 397 (should not be protected. = 0;
433 (i > j): 398 (i > j):
434 (not necessary, but looks 0) 399 (not necessary, but looks 0)
435 400
436 The default values of lowmem_reserve_ratio[i] 401 The default values of lowmem_reserve_ratio[i] are
437 402
438 === ==================================== 403 === ====================================
439 256 (if zone[i] means DMA or DMA32 zone) 404 256 (if zone[i] means DMA or DMA32 zone)
440 32 (others) 405 32 (others)
441 === ==================================== 406 === ====================================
442 407
443 As above expression, they are reciprocal numbe 408 As above expression, they are reciprocal number of ratio.
444 256 means 1/256. # of protection pages becomes 409 256 means 1/256. # of protection pages becomes about "0.39%" of total managed
445 pages of higher zones on the node. 410 pages of higher zones on the node.
446 411
447 If you would like to protect more pages, small 412 If you would like to protect more pages, smaller values are effective.
448 The minimum value is 1 (1/1 -> 100%). The valu 413 The minimum value is 1 (1/1 -> 100%). The value less than 1 completely
449 disables protection of the pages. 414 disables protection of the pages.
450 415
451 416
452 max_map_count: 417 max_map_count:
453 ============== 418 ==============
454 419
455 This file contains the maximum number of memor 420 This file contains the maximum number of memory map areas a process
456 may have. Memory map areas are used as a side- 421 may have. Memory map areas are used as a side-effect of calling
457 malloc, directly by mmap, mprotect, and madvis 422 malloc, directly by mmap, mprotect, and madvise, and also when loading
458 shared libraries. 423 shared libraries.
459 424
460 While most applications need less than a thous 425 While most applications need less than a thousand maps, certain
461 programs, particularly malloc debuggers, may c 426 programs, particularly malloc debuggers, may consume lots of them,
462 e.g., up to one or two maps per allocation. 427 e.g., up to one or two maps per allocation.
463 428
464 The default value is 65530. !! 429 The default value is 65536.
465 <<
466 <<
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_ <<
480 430
481 431
482 memory_failure_early_kill: 432 memory_failure_early_kill:
483 ========================== 433 ==========================
484 434
485 Control how to kill processes when uncorrected 435 Control how to kill processes when uncorrected memory error (typically
486 a 2bit error in a memory module) is detected i 436 a 2bit error in a memory module) is detected in the background by hardware
487 that cannot be handled by the kernel. In some 437 that cannot be handled by the kernel. In some cases (like the page
488 still having a valid copy on disk) the kernel 438 still having a valid copy on disk) the kernel will handle the failure
489 transparently without affecting any applicatio 439 transparently without affecting any applications. But if there is
490 no other up-to-date copy of the data it will k !! 440 no other uptodate copy of the data it will kill to prevent any data
491 corruptions from propagating. 441 corruptions from propagating.
492 442
493 1: Kill all processes that have the corrupted 443 1: Kill all processes that have the corrupted and not reloadable page mapped
494 as soon as the corruption is detected. Note t 444 as soon as the corruption is detected. Note this is not supported
495 for a few types of pages, like kernel internal 445 for a few types of pages, like kernel internally allocated data or
496 the swap cache, but works for the majority of 446 the swap cache, but works for the majority of user pages.
497 447
498 0: Only unmap the corrupted page from all proc 448 0: Only unmap the corrupted page from all processes and only kill a process
499 who tries to access it. 449 who tries to access it.
500 450
501 The kill is done using a catchable SIGBUS with 451 The kill is done using a catchable SIGBUS with BUS_MCEERR_AO, so processes can
502 handle this if they want to. 452 handle this if they want to.
503 453
504 This is only active on architectures/platforms 454 This is only active on architectures/platforms with advanced machine
505 check handling and depends on the hardware cap 455 check handling and depends on the hardware capabilities.
506 456
507 Applications can override this setting individ 457 Applications can override this setting individually with the PR_MCE_KILL prctl
508 458
509 459
510 memory_failure_recovery 460 memory_failure_recovery
511 ======================= 461 =======================
512 462
513 Enable memory failure recovery (when supported 463 Enable memory failure recovery (when supported by the platform)
514 464
515 1: Attempt recovery. 465 1: Attempt recovery.
516 466
517 0: Always panic on a memory failure. 467 0: Always panic on a memory failure.
518 468
519 469
520 min_free_kbytes 470 min_free_kbytes
521 =============== 471 ===============
522 472
523 This is used to force the Linux VM to keep a m 473 This is used to force the Linux VM to keep a minimum number
524 of kilobytes free. The VM uses this number to 474 of kilobytes free. The VM uses this number to compute a
525 watermark[WMARK_MIN] value for each lowmem zon 475 watermark[WMARK_MIN] value for each lowmem zone in the system.
526 Each lowmem zone gets a number of reserved fre 476 Each lowmem zone gets a number of reserved free pages based
527 proportionally on its size. 477 proportionally on its size.
528 478
529 Some minimal amount of memory is needed to sat 479 Some minimal amount of memory is needed to satisfy PF_MEMALLOC
530 allocations; if you set this to lower than 102 480 allocations; if you set this to lower than 1024KB, your system will
531 become subtly broken, and prone to deadlock un 481 become subtly broken, and prone to deadlock under high loads.
532 482
533 Setting this too high will OOM your machine in 483 Setting this too high will OOM your machine instantly.
534 484
535 485
536 min_slab_ratio 486 min_slab_ratio
537 ============== 487 ==============
538 488
539 This is available only on NUMA kernels. 489 This is available only on NUMA kernels.
540 490
541 A percentage of the total pages in each zone. 491 A percentage of the total pages in each zone. On Zone reclaim
542 (fallback from the local zone occurs) slabs wi 492 (fallback from the local zone occurs) slabs will be reclaimed if more
543 than this percentage of pages in a zone are re 493 than this percentage of pages in a zone are reclaimable slab pages.
544 This insures that the slab growth stays under 494 This insures that the slab growth stays under control even in NUMA
545 systems that rarely perform global reclaim. 495 systems that rarely perform global reclaim.
546 496
547 The default is 5 percent. 497 The default is 5 percent.
548 498
549 Note that slab reclaim is triggered in a per z 499 Note that slab reclaim is triggered in a per zone / node fashion.
550 The process of reclaiming slab memory is curre 500 The process of reclaiming slab memory is currently not node specific
551 and may not be fast. 501 and may not be fast.
552 502
553 503
554 min_unmapped_ratio 504 min_unmapped_ratio
555 ================== 505 ==================
556 506
557 This is available only on NUMA kernels. 507 This is available only on NUMA kernels.
558 508
559 This is a percentage of the total pages in eac 509 This is a percentage of the total pages in each zone. Zone reclaim will
560 only occur if more than this percentage of pag 510 only occur if more than this percentage of pages are in a state that
561 zone_reclaim_mode allows to be reclaimed. 511 zone_reclaim_mode allows to be reclaimed.
562 512
563 If zone_reclaim_mode has the value 4 OR'd, the 513 If zone_reclaim_mode has the value 4 OR'd, then the percentage is compared
564 against all file-backed unmapped pages includi 514 against all file-backed unmapped pages including swapcache pages and tmpfs
565 files. Otherwise, only unmapped pages backed b 515 files. Otherwise, only unmapped pages backed by normal files but not tmpfs
566 files and similar are considered. 516 files and similar are considered.
567 517
568 The default is 1 percent. 518 The default is 1 percent.
569 519
570 520
571 mmap_min_addr 521 mmap_min_addr
572 ============= 522 =============
573 523
574 This file indicates the amount of address spac 524 This file indicates the amount of address space which a user process will
575 be restricted from mmapping. Since kernel nul 525 be restricted from mmapping. Since kernel null dereference bugs could
576 accidentally operate based on the information 526 accidentally operate based on the information in the first couple of pages
577 of memory userspace processes should not be al 527 of memory userspace processes should not be allowed to write to them. By
578 default this value is set to 0 and no protecti 528 default this value is set to 0 and no protections will be enforced by the
579 security module. Setting this value to someth 529 security module. Setting this value to something like 64k will allow the
580 vast majority of applications to work correctl 530 vast majority of applications to work correctly and provide defense in depth
581 against future potential kernel bugs. 531 against future potential kernel bugs.
582 532
583 533
584 mmap_rnd_bits 534 mmap_rnd_bits
585 ============= 535 =============
586 536
587 This value can be used to select the number of 537 This value can be used to select the number of bits to use to
588 determine the random offset to the base addres 538 determine the random offset to the base address of vma regions
589 resulting from mmap allocations on architectur 539 resulting from mmap allocations on architectures which support
590 tuning address space randomization. This valu 540 tuning address space randomization. This value will be bounded
591 by the architecture's minimum and maximum supp 541 by the architecture's minimum and maximum supported values.
592 542
593 This value can be changed after boot using the 543 This value can be changed after boot using the
594 /proc/sys/vm/mmap_rnd_bits tunable 544 /proc/sys/vm/mmap_rnd_bits tunable
595 545
596 546
597 mmap_rnd_compat_bits 547 mmap_rnd_compat_bits
598 ==================== 548 ====================
599 549
600 This value can be used to select the number of 550 This value can be used to select the number of bits to use to
601 determine the random offset to the base addres 551 determine the random offset to the base address of vma regions
602 resulting from mmap allocations for applicatio 552 resulting from mmap allocations for applications run in
603 compatibility mode on architectures which supp 553 compatibility mode on architectures which support tuning address
604 space randomization. This value will be bound 554 space randomization. This value will be bounded by the
605 architecture's minimum and maximum supported v 555 architecture's minimum and maximum supported values.
606 556
607 This value can be changed after boot using the 557 This value can be changed after boot using the
608 /proc/sys/vm/mmap_rnd_compat_bits tunable 558 /proc/sys/vm/mmap_rnd_compat_bits tunable
609 559
610 560
611 nr_hugepages 561 nr_hugepages
612 ============ 562 ============
613 563
614 Change the minimum size of the hugepage pool. 564 Change the minimum size of the hugepage pool.
615 565
616 See Documentation/admin-guide/mm/hugetlbpage.r 566 See Documentation/admin-guide/mm/hugetlbpage.rst
617 567
618 568
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 569 nr_hugepages_mempolicy
657 ====================== 570 ======================
658 571
659 Change the size of the hugepage pool at run-ti 572 Change the size of the hugepage pool at run-time on a specific
660 set of NUMA nodes. 573 set of NUMA nodes.
661 574
662 See Documentation/admin-guide/mm/hugetlbpage.r 575 See Documentation/admin-guide/mm/hugetlbpage.rst
663 576
664 577
665 nr_overcommit_hugepages 578 nr_overcommit_hugepages
666 ======================= 579 =======================
667 580
668 Change the maximum size of the hugepage pool. 581 Change the maximum size of the hugepage pool. The maximum is
669 nr_hugepages + nr_overcommit_hugepages. 582 nr_hugepages + nr_overcommit_hugepages.
670 583
671 See Documentation/admin-guide/mm/hugetlbpage.r 584 See Documentation/admin-guide/mm/hugetlbpage.rst
672 585
673 586
674 nr_trim_pages 587 nr_trim_pages
675 ============= 588 =============
676 589
677 This is available only on NOMMU kernels. 590 This is available only on NOMMU kernels.
678 591
679 This value adjusts the excess page trimming be 592 This value adjusts the excess page trimming behaviour of power-of-2 aligned
680 NOMMU mmap allocations. 593 NOMMU mmap allocations.
681 594
682 A value of 0 disables trimming of allocations 595 A value of 0 disables trimming of allocations entirely, while a value of 1
683 trims excess pages aggressively. Any value >= 596 trims excess pages aggressively. Any value >= 1 acts as the watermark where
684 trimming of allocations is initiated. 597 trimming of allocations is initiated.
685 598
686 The default value is 1. 599 The default value is 1.
687 600
688 See Documentation/admin-guide/mm/nommu-mmap.rs 601 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
689 602
690 603
691 numa_zonelist_order 604 numa_zonelist_order
692 =================== 605 ===================
693 606
694 This sysctl is only for NUMA and it is depreca 607 This sysctl is only for NUMA and it is deprecated. Anything but
695 Node order will fail! 608 Node order will fail!
696 609
697 'where the memory is allocated from' is contro 610 'where the memory is allocated from' is controlled by zonelists.
698 611
699 (This documentation ignores ZONE_HIGHMEM/ZONE_ 612 (This documentation ignores ZONE_HIGHMEM/ZONE_DMA32 for simple explanation.
700 you may be able to read ZONE_DMA as ZONE_DMA32 613 you may be able to read ZONE_DMA as ZONE_DMA32...)
701 614
702 In non-NUMA case, a zonelist for GFP_KERNEL is 615 In non-NUMA case, a zonelist for GFP_KERNEL is ordered as following.
703 ZONE_NORMAL -> ZONE_DMA 616 ZONE_NORMAL -> ZONE_DMA
704 This means that a memory allocation request fo 617 This means that a memory allocation request for GFP_KERNEL will
705 get memory from ZONE_DMA only when ZONE_NORMAL 618 get memory from ZONE_DMA only when ZONE_NORMAL is not available.
706 619
707 In NUMA case, you can think of following 2 typ 620 In NUMA case, you can think of following 2 types of order.
708 Assume 2 node NUMA and below is zonelist of No 621 Assume 2 node NUMA and below is zonelist of Node(0)'s GFP_KERNEL::
709 622
710 (A) Node(0) ZONE_NORMAL -> Node(0) ZONE_DMA 623 (A) Node(0) ZONE_NORMAL -> Node(0) ZONE_DMA -> Node(1) ZONE_NORMAL
711 (B) Node(0) ZONE_NORMAL -> Node(1) ZONE_NORM 624 (B) Node(0) ZONE_NORMAL -> Node(1) ZONE_NORMAL -> Node(0) ZONE_DMA.
712 625
713 Type(A) offers the best locality for processes 626 Type(A) offers the best locality for processes on Node(0), but ZONE_DMA
714 will be used before ZONE_NORMAL exhaustion. Th 627 will be used before ZONE_NORMAL exhaustion. This increases possibility of
715 out-of-memory(OOM) of ZONE_DMA because ZONE_DM 628 out-of-memory(OOM) of ZONE_DMA because ZONE_DMA is tend to be small.
716 629
717 Type(B) cannot offer the best locality but is 630 Type(B) cannot offer the best locality but is more robust against OOM of
718 the DMA zone. 631 the DMA zone.
719 632
720 Type(A) is called as "Node" order. Type (B) is 633 Type(A) is called as "Node" order. Type (B) is "Zone" order.
721 634
722 "Node order" orders the zonelists by node, the 635 "Node order" orders the zonelists by node, then by zone within each node.
723 Specify "[Nn]ode" for node order 636 Specify "[Nn]ode" for node order
724 637
725 "Zone Order" orders the zonelists by zone type 638 "Zone Order" orders the zonelists by zone type, then by node within each
726 zone. Specify "[Zz]one" for zone order. 639 zone. Specify "[Zz]one" for zone order.
727 640
728 Specify "[Dd]efault" to request automatic conf 641 Specify "[Dd]efault" to request automatic configuration.
729 642
730 On 32-bit, the Normal zone needs to be preserv 643 On 32-bit, the Normal zone needs to be preserved for allocations accessible
731 by the kernel, so "zone" order will be selecte 644 by the kernel, so "zone" order will be selected.
732 645
733 On 64-bit, devices that require DMA32/DMA are 646 On 64-bit, devices that require DMA32/DMA are relatively rare, so "node"
734 order will be selected. 647 order will be selected.
735 648
736 Default order is recommended unless this is ca 649 Default order is recommended unless this is causing problems for your
737 system/application. 650 system/application.
738 651
739 652
740 oom_dump_tasks 653 oom_dump_tasks
741 ============== 654 ==============
742 655
743 Enables a system-wide task dump (excluding ker 656 Enables a system-wide task dump (excluding kernel threads) to be produced
744 when the kernel performs an OOM-killing and in 657 when the kernel performs an OOM-killing and includes such information as
745 pid, uid, tgid, vm size, rss, pgtables_bytes, 658 pid, uid, tgid, vm size, rss, pgtables_bytes, swapents, oom_score_adj
746 score, and name. This is helpful to determine 659 score, and name. This is helpful to determine why the OOM killer was
747 invoked, to identify the rogue task that cause 660 invoked, to identify the rogue task that caused it, and to determine why
748 the OOM killer chose the task it did to kill. 661 the OOM killer chose the task it did to kill.
749 662
750 If this is set to zero, this information is su 663 If this is set to zero, this information is suppressed. On very
751 large systems with thousands of tasks it may n 664 large systems with thousands of tasks it may not be feasible to dump
752 the memory state information for each one. Su 665 the memory state information for each one. Such systems should not
753 be forced to incur a performance penalty in OO 666 be forced to incur a performance penalty in OOM conditions when the
754 information may not be desired. 667 information may not be desired.
755 668
756 If this is set to non-zero, this information i 669 If this is set to non-zero, this information is shown whenever the
757 OOM killer actually kills a memory-hogging tas 670 OOM killer actually kills a memory-hogging task.
758 671
759 The default value is 1 (enabled). 672 The default value is 1 (enabled).
760 673
761 674
762 oom_kill_allocating_task 675 oom_kill_allocating_task
763 ======================== 676 ========================
764 677
765 This enables or disables killing the OOM-trigg 678 This enables or disables killing the OOM-triggering task in
766 out-of-memory situations. 679 out-of-memory situations.
767 680
768 If this is set to zero, the OOM killer will sc 681 If this is set to zero, the OOM killer will scan through the entire
769 tasklist and select a task based on heuristics 682 tasklist and select a task based on heuristics to kill. This normally
770 selects a rogue memory-hogging task that frees 683 selects a rogue memory-hogging task that frees up a large amount of
771 memory when killed. 684 memory when killed.
772 685
773 If this is set to non-zero, the OOM killer sim 686 If this is set to non-zero, the OOM killer simply kills the task that
774 triggered the out-of-memory condition. This a 687 triggered the out-of-memory condition. This avoids the expensive
775 tasklist scan. 688 tasklist scan.
776 689
777 If panic_on_oom is selected, it takes preceden 690 If panic_on_oom is selected, it takes precedence over whatever value
778 is used in oom_kill_allocating_task. 691 is used in oom_kill_allocating_task.
779 692
780 The default value is 0. 693 The default value is 0.
781 694
782 695
783 overcommit_kbytes 696 overcommit_kbytes
784 ================= 697 =================
785 698
786 When overcommit_memory is set to 2, the commit 699 When overcommit_memory is set to 2, the committed address space is not
787 permitted to exceed swap plus this amount of p 700 permitted to exceed swap plus this amount of physical RAM. See below.
788 701
789 Note: overcommit_kbytes is the counterpart of 702 Note: overcommit_kbytes is the counterpart of overcommit_ratio. Only one
790 of them may be specified at a time. Setting on 703 of them may be specified at a time. Setting one disables the other (which
791 then appears as 0 when read). 704 then appears as 0 when read).
792 705
793 706
794 overcommit_memory 707 overcommit_memory
795 ================= 708 =================
796 709
797 This value contains a flag that enables memory 710 This value contains a flag that enables memory overcommitment.
798 711
799 When this flag is 0, the kernel compares the u !! 712 When this flag is 0, the kernel attempts to estimate the amount
800 size against total memory plus swap and reject !! 713 of free memory left when userspace requests more memory.
801 714
802 When this flag is 1, the kernel pretends there 715 When this flag is 1, the kernel pretends there is always enough
803 memory until it actually runs out. 716 memory until it actually runs out.
804 717
805 When this flag is 2, the kernel uses a "never 718 When this flag is 2, the kernel uses a "never overcommit"
806 policy that attempts to prevent any overcommit 719 policy that attempts to prevent any overcommit of memory.
807 Note that user_reserve_kbytes affects this pol 720 Note that user_reserve_kbytes affects this policy.
808 721
809 This feature can be very useful because there 722 This feature can be very useful because there are a lot of
810 programs that malloc() huge amounts of memory 723 programs that malloc() huge amounts of memory "just-in-case"
811 and don't use much of it. 724 and don't use much of it.
812 725
813 The default value is 0. 726 The default value is 0.
814 727
815 See Documentation/mm/overcommit-accounting.rst !! 728 See Documentation/vm/overcommit-accounting.rst and
816 mm/util.c::__vm_enough_memory() for more infor 729 mm/util.c::__vm_enough_memory() for more information.
817 730
818 731
819 overcommit_ratio 732 overcommit_ratio
820 ================ 733 ================
821 734
822 When overcommit_memory is set to 2, the commit 735 When overcommit_memory is set to 2, the committed address
823 space is not permitted to exceed swap plus thi 736 space is not permitted to exceed swap plus this percentage
824 of physical RAM. See above. 737 of physical RAM. See above.
825 738
826 739
827 page-cluster 740 page-cluster
828 ============ 741 ============
829 742
830 page-cluster controls the number of pages up t 743 page-cluster controls the number of pages up to which consecutive pages
831 are read in from swap in a single attempt. Thi 744 are read in from swap in a single attempt. This is the swap counterpart
832 to page cache readahead. 745 to page cache readahead.
833 The mentioned consecutivity is not in terms of 746 The mentioned consecutivity is not in terms of virtual/physical addresses,
834 but consecutive on swap space - that means the 747 but consecutive on swap space - that means they were swapped out together.
835 748
836 It is a logarithmic value - setting it to zero 749 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 750 it to 1 means "2 pages", setting it to 2 means "4 pages", etc.
838 Zero disables swap readahead completely. 751 Zero disables swap readahead completely.
839 752
840 The default value is three (eight pages at a t 753 The default value is three (eight pages at a time). There may be some
841 small benefits in tuning this to a different v 754 small benefits in tuning this to a different value if your workload is
842 swap-intensive. 755 swap-intensive.
843 756
844 Lower values mean lower latencies for initial 757 Lower values mean lower latencies for initial faults, but at the same time
845 extra faults and I/O delays for following faul 758 extra faults and I/O delays for following faults if they would have been part of
846 that consecutive pages readahead would have br 759 that consecutive pages readahead would have brought in.
847 760
848 761
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 762 panic_on_oom
858 ============ 763 ============
859 764
860 This enables or disables panic on out-of-memor 765 This enables or disables panic on out-of-memory feature.
861 766
862 If this is set to 0, the kernel will kill some 767 If this is set to 0, the kernel will kill some rogue process,
863 called oom_killer. Usually, oom_killer can ki 768 called oom_killer. Usually, oom_killer can kill rogue processes and
864 system will survive. 769 system will survive.
865 770
866 If this is set to 1, the kernel panics when ou 771 If this is set to 1, the kernel panics when out-of-memory happens.
867 However, if a process limits using nodes by me 772 However, if a process limits using nodes by mempolicy/cpusets,
868 and those nodes become memory exhaustion statu 773 and those nodes become memory exhaustion status, one process
869 may be killed by oom-killer. No panic occurs i 774 may be killed by oom-killer. No panic occurs in this case.
870 Because other nodes' memory may be free. This 775 Because other nodes' memory may be free. This means system total status
871 may be not fatal yet. 776 may be not fatal yet.
872 777
873 If this is set to 2, the kernel panics compuls 778 If this is set to 2, the kernel panics compulsorily even on the
874 above-mentioned. Even oom happens under memory 779 above-mentioned. Even oom happens under memory cgroup, the whole
875 system panics. 780 system panics.
876 781
877 The default value is 0. 782 The default value is 0.
878 783
879 1 and 2 are for failover of clustering. Please 784 1 and 2 are for failover of clustering. Please select either
880 according to your policy of failover. 785 according to your policy of failover.
881 786
882 panic_on_oom=2+kdump gives you very strong too 787 panic_on_oom=2+kdump gives you very strong tool to investigate
883 why oom happens. You can get snapshot. 788 why oom happens. You can get snapshot.
884 789
885 790
886 percpu_pagelist_high_fraction !! 791 percpu_pagelist_fraction
887 ============================= !! 792 ========================
888 793
889 This is the fraction of pages in each zone tha !! 794 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 !! 795 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 !! 796 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 !! 797 allocated in any single per_cpu_pagelist. This entry only changes the value
893 on per-cpu page lists. This entry only changes !! 798 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 !! 799 1/100th of each zone to each per cpu page list.
895 each zone between per-cpu lists. !! 800
896 !! 801 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 !! 802 set to pcp->high/4. The upper limit of batch is (PAGE_SHIFT * 8)
898 the value of the high fraction so allocation l !! 803
899 !! 804 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 !! 805 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 !! 806 sysctl, it will revert to this default behavior.
902 online CPUs. If the user writes '0' to this s <<
903 this default behavior. <<
904 807
905 808
906 stat_interval 809 stat_interval
907 ============= 810 =============
908 811
909 The time interval between which vm statistics 812 The time interval between which vm statistics are updated. The default
910 is 1 second. 813 is 1 second.
911 814
912 815
913 stat_refresh 816 stat_refresh
914 ============ 817 ============
915 818
916 Any read or write (by root only) flushes all t 819 Any read or write (by root only) flushes all the per-cpu vm statistics
917 into their global totals, for more accurate re 820 into their global totals, for more accurate reports when testing
918 e.g. cat /proc/sys/vm/stat_refresh /proc/memin 821 e.g. cat /proc/sys/vm/stat_refresh /proc/meminfo
919 822
920 As a side-effect, it also checks for negative 823 As a side-effect, it also checks for negative totals (elsewhere reported
921 as 0) and "fails" with EINVAL if any are found 824 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 825 (At time of writing, a few stats are known sometimes to be found negative,
923 with no ill effects: errors and warnings on th 826 with no ill effects: errors and warnings on these stats are suppressed.)
924 827
925 828
926 numa_stat 829 numa_stat
927 ========= 830 =========
928 831
929 This interface allows runtime configuration of 832 This interface allows runtime configuration of numa statistics.
930 833
931 When page allocation performance becomes a bot 834 When page allocation performance becomes a bottleneck and you can tolerate
932 some possible tool breakage and decreased numa 835 some possible tool breakage and decreased numa counter precision, you can
933 do:: 836 do::
934 837
935 echo 0 > /proc/sys/vm/numa_stat 838 echo 0 > /proc/sys/vm/numa_stat
936 839
937 When page allocation performance is not a bott 840 When page allocation performance is not a bottleneck and you want all
938 tooling to work, you can do:: 841 tooling to work, you can do::
939 842
940 echo 1 > /proc/sys/vm/numa_stat 843 echo 1 > /proc/sys/vm/numa_stat
941 844
942 845
943 swappiness 846 swappiness
944 ========== 847 ==========
945 848
946 This control is used to define the rough relat 849 This control is used to define the rough relative IO cost of swapping
947 and filesystem paging, as a value between 0 an 850 and filesystem paging, as a value between 0 and 200. At 100, the VM
948 assumes equal IO cost and will thus apply memo 851 assumes equal IO cost and will thus apply memory pressure to the page
949 cache and swap-backed pages equally; lower val 852 cache and swap-backed pages equally; lower values signify more
950 expensive swap IO, higher values indicates che 853 expensive swap IO, higher values indicates cheaper.
951 854
952 Keep in mind that filesystem IO patterns under 855 Keep in mind that filesystem IO patterns under memory pressure tend to
953 be more efficient than swap's random IO. An op 856 be more efficient than swap's random IO. An optimal value will require
954 experimentation and will also be workload-depe 857 experimentation and will also be workload-dependent.
955 858
956 The default value is 60. 859 The default value is 60.
957 860
958 For in-memory swap, like zram or zswap, as wel 861 For in-memory swap, like zram or zswap, as well as hybrid setups that
959 have swap on faster devices than the filesyste 862 have swap on faster devices than the filesystem, values beyond 100 can
960 be considered. For example, if the random IO a 863 be considered. For example, if the random IO against the swap device
961 is on average 2x faster than IO from the files 864 is on average 2x faster than IO from the filesystem, swappiness should
962 be 133 (x + 2x = 200, 2x = 133.33). 865 be 133 (x + 2x = 200, 2x = 133.33).
963 866
964 At 0, the kernel will not initiate swap until 867 At 0, the kernel will not initiate swap until the amount of free and
965 file-backed pages is less than the high waterm 868 file-backed pages is less than the high watermark in a zone.
966 869
967 870
968 unprivileged_userfaultfd 871 unprivileged_userfaultfd
969 ======================== 872 ========================
970 873
971 This flag controls the mode in which unprivile !! 874 This flag controls whether unprivileged users can use the userfaultfd
972 userfaultfd system calls. Set this to 0 to res !! 875 system calls. Set this to 1 to allow unprivileged users to use the
973 to handle page faults in user mode only. In th !! 876 userfaultfd system calls, or set this to 0 to restrict userfaultfd to only
974 SYS_CAP_PTRACE must pass UFFD_USER_MODE_ONLY i !! 877 privileged users (with SYS_CAP_PTRACE capability).
975 succeed. Prohibiting use of userfaultfd for ha <<
976 mode may make certain vulnerabilities more dif <<
977 <<
978 Set this to 1 to allow unprivileged users to u <<
979 calls without any restrictions. <<
980 878
981 The default value is 0. !! 879 The default value is 1.
982 880
983 Another way to control permissions for userfau <<
984 /dev/userfaultfd instead of userfaultfd(2). Se <<
985 Documentation/admin-guide/mm/userfaultfd.rst. <<
986 881
987 user_reserve_kbytes 882 user_reserve_kbytes
988 =================== 883 ===================
989 884
990 When overcommit_memory is set to 2, "never ove 885 When overcommit_memory is set to 2, "never overcommit" mode, reserve
991 min(3% of current process size, user_reserve_k 886 min(3% of current process size, user_reserve_kbytes) of free memory.
992 This is intended to prevent a user from starti 887 This is intended to prevent a user from starting a single memory hogging
993 process, such that they cannot recover (kill t 888 process, such that they cannot recover (kill the hog).
994 889
995 user_reserve_kbytes defaults to min(3% of the 890 user_reserve_kbytes defaults to min(3% of the current process size, 128MB).
996 891
997 If this is reduced to zero, then the user will 892 If this is reduced to zero, then the user will be allowed to allocate
998 all free memory with a single process, minus a 893 all free memory with a single process, minus admin_reserve_kbytes.
999 Any subsequent attempts to execute a command w 894 Any subsequent attempts to execute a command will result in
1000 "fork: Cannot allocate memory". 895 "fork: Cannot allocate memory".
1001 896
1002 Changing this takes effect whenever an applic 897 Changing this takes effect whenever an application requests memory.
1003 898
1004 899
1005 vfs_cache_pressure 900 vfs_cache_pressure
1006 ================== 901 ==================
1007 902
1008 This percentage value controls the tendency o 903 This percentage value controls the tendency of the kernel to reclaim
1009 the memory which is used for caching of direc 904 the memory which is used for caching of directory and inode objects.
1010 905
1011 At the default value of vfs_cache_pressure=10 906 At the default value of vfs_cache_pressure=100 the kernel will attempt to
1012 reclaim dentries and inodes at a "fair" rate 907 reclaim dentries and inodes at a "fair" rate with respect to pagecache and
1013 swapcache reclaim. Decreasing vfs_cache_pres 908 swapcache reclaim. Decreasing vfs_cache_pressure causes the kernel to prefer
1014 to retain dentry and inode caches. When vfs_c 909 to retain dentry and inode caches. When vfs_cache_pressure=0, the kernel will
1015 never reclaim dentries and inodes due to memo 910 never reclaim dentries and inodes due to memory pressure and this can easily
1016 lead to out-of-memory conditions. Increasing 911 lead to out-of-memory conditions. Increasing vfs_cache_pressure beyond 100
1017 causes the kernel to prefer to reclaim dentri 912 causes the kernel to prefer to reclaim dentries and inodes.
1018 913
1019 Increasing vfs_cache_pressure significantly b 914 Increasing vfs_cache_pressure significantly beyond 100 may have negative
1020 performance impact. Reclaim code needs to tak 915 performance impact. Reclaim code needs to take various locks to find freeable
1021 directory and inode objects. With vfs_cache_p 916 directory and inode objects. With vfs_cache_pressure=1000, it will look for
1022 ten times more freeable objects than there ar 917 ten times more freeable objects than there are.
1023 918
1024 919
1025 watermark_boost_factor 920 watermark_boost_factor
1026 ====================== 921 ======================
1027 922
1028 This factor controls the level of reclaim whe 923 This factor controls the level of reclaim when memory is being fragmented.
1029 It defines the percentage of the high waterma 924 It defines the percentage of the high watermark of a zone that will be
1030 reclaimed if pages of different mobility are 925 reclaimed if pages of different mobility are being mixed within pageblocks.
1031 The intent is that compaction has less work t 926 The intent is that compaction has less work to do in the future and to
1032 increase the success rate of future high-orde 927 increase the success rate of future high-order allocations such as SLUB
1033 allocations, THP and hugetlbfs pages. 928 allocations, THP and hugetlbfs pages.
1034 929
1035 To make it sensible with respect to the water 930 To make it sensible with respect to the watermark_scale_factor
1036 parameter, the unit is in fractions of 10,000 931 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 !! 932 15,000 on !DISCONTIGMEM configurations means that up to 150% of the high
1038 event of a pageblock being mixed due to fragm !! 933 watermark will be reclaimed in the event of a pageblock being mixed due
1039 is determined by the number of fragmentation !! 934 to fragmentation. The level of reclaim is determined by the number of
1040 recent past. If this value is smaller than a !! 935 fragmentation events that occurred in the recent past. If this value is
1041 worth of pages will be reclaimed (e.g. 2MB o !! 936 smaller than a pageblock then a pageblocks worth of pages will be reclaimed
1042 of 0 will disable the feature. !! 937 (e.g. 2MB on 64-bit x86). A boost factor of 0 will disable the feature.
1043 938
1044 939
1045 watermark_scale_factor 940 watermark_scale_factor
1046 ====================== 941 ======================
1047 942
1048 This factor controls the aggressiveness of ks 943 This factor controls the aggressiveness of kswapd. It defines the
1049 amount of memory left in a node/system before 944 amount of memory left in a node/system before kswapd is woken up and
1050 how much memory needs to be free before kswap 945 how much memory needs to be free before kswapd goes back to sleep.
1051 946
1052 The unit is in fractions of 10,000. The defau 947 The unit is in fractions of 10,000. The default value of 10 means the
1053 distances between watermarks are 0.1% of the 948 distances between watermarks are 0.1% of the available memory in the
1054 node/system. The maximum value is 3000, or 30 !! 949 node/system. The maximum value is 1000, or 10% of memory.
1055 950
1056 A high rate of threads entering direct reclai 951 A high rate of threads entering direct reclaim (allocstall) or kswapd
1057 going to sleep prematurely (kswapd_low_wmark_ 952 going to sleep prematurely (kswapd_low_wmark_hit_quickly) can indicate
1058 that the number of free pages kswapd maintain 953 that the number of free pages kswapd maintains for latency reasons is
1059 too small for the allocation bursts occurring 954 too small for the allocation bursts occurring in the system. This knob
1060 can then be used to tune kswapd aggressivenes 955 can then be used to tune kswapd aggressiveness accordingly.
1061 956
1062 957
1063 zone_reclaim_mode 958 zone_reclaim_mode
1064 ================= 959 =================
1065 960
1066 Zone_reclaim_mode allows someone to set more 961 Zone_reclaim_mode allows someone to set more or less aggressive approaches to
1067 reclaim memory when a zone runs out of memory 962 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 963 zone reclaim occurs. Allocations will be satisfied from other zones / nodes
1069 in the system. 964 in the system.
1070 965
1071 This is value OR'ed together of 966 This is value OR'ed together of
1072 967
1073 = =================================== 968 = ===================================
1074 1 Zone reclaim on 969 1 Zone reclaim on
1075 2 Zone reclaim writes dirty pages out 970 2 Zone reclaim writes dirty pages out
1076 4 Zone reclaim swaps pages 971 4 Zone reclaim swaps pages
1077 = =================================== 972 = ===================================
1078 973
1079 zone_reclaim_mode is disabled by default. Fo 974 zone_reclaim_mode is disabled by default. For file servers or workloads
1080 that benefit from having their data cached, z 975 that benefit from having their data cached, zone_reclaim_mode should be
1081 left disabled as the caching effect is likely 976 left disabled as the caching effect is likely to be more important than
1082 data locality. 977 data locality.
1083 978
1084 Consider enabling one or more zone_reclaim mo !! 979 zone_reclaim may be enabled if it's known that the workload is partitioned
1085 workload is partitioned such that each partit !! 980 such that each partition fits within a NUMA node and that accessing remote
1086 and that accessing remote memory would cause !! 981 memory would cause a measurable performance reduction. The page allocator
1087 reduction. The page allocator will take addi !! 982 will then reclaim easily reusable pages (those page cache pages that are
1088 allocating off node pages. !! 983 currently not used) before allocating off node pages.
1089 984
1090 Allowing zone reclaim to write out pages stop 985 Allowing zone reclaim to write out pages stops processes that are
1091 writing large amounts of data from dirtying p 986 writing large amounts of data from dirtying pages on other nodes. Zone
1092 reclaim will write out dirty pages if a zone 987 reclaim will write out dirty pages if a zone fills up and so effectively
1093 throttle the process. This may decrease the p 988 throttle the process. This may decrease the performance of a single process
1094 since it cannot use all of system memory to b 989 since it cannot use all of system memory to buffer the outgoing writes
1095 anymore but it preserve the memory on other n 990 anymore but it preserve the memory on other nodes so that the performance
1096 of other processes running on other nodes wil 991 of other processes running on other nodes will not be affected.
1097 992
1098 Allowing regular swap effectively restricts a 993 Allowing regular swap effectively restricts allocations to the local
1099 node unless explicitly overridden by memory p 994 node unless explicitly overridden by memory policies or cpuset
1100 configurations. 995 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.