]>
Commit | Line | Data |
---|---|---|
db0fb184 | 1 | Documentation for /proc/sys/vm/* kernel version 2.6.29 |
1da177e4 | 2 | (c) 1998, 1999, Rik van Riel <riel@nl.linux.org> |
db0fb184 | 3 | (c) 2008 Peter W. Morreale <pmorreale@novell.com> |
1da177e4 LT |
4 | |
5 | For general info and legal blurb, please look in README. | |
6 | ||
7 | ============================================================== | |
8 | ||
9 | This file contains the documentation for the sysctl files in | |
db0fb184 | 10 | /proc/sys/vm and is valid for Linux kernel version 2.6.29. |
1da177e4 LT |
11 | |
12 | The files in this directory can be used to tune the operation | |
13 | of the virtual memory (VM) subsystem of the Linux kernel and | |
14 | the writeout of dirty data to disk. | |
15 | ||
16 | Default values and initialization routines for most of these | |
17 | files can be found in mm/swap.c. | |
18 | ||
19 | Currently, these files are in /proc/sys/vm: | |
db0fb184 | 20 | |
4eeab4f5 | 21 | - admin_reserve_kbytes |
db0fb184 | 22 | - block_dump |
76ab0f53 | 23 | - compact_memory |
db0fb184 | 24 | - dirty_background_bytes |
1da177e4 | 25 | - dirty_background_ratio |
db0fb184 | 26 | - dirty_bytes |
1da177e4 | 27 | - dirty_expire_centisecs |
db0fb184 | 28 | - dirty_ratio |
1da177e4 | 29 | - dirty_writeback_centisecs |
db0fb184 | 30 | - drop_caches |
5e771905 | 31 | - extfrag_threshold |
db0fb184 PM |
32 | - hugepages_treat_as_movable |
33 | - hugetlb_shm_group | |
34 | - laptop_mode | |
35 | - legacy_va_layout | |
36 | - lowmem_reserve_ratio | |
1da177e4 | 37 | - max_map_count |
6a46079c AK |
38 | - memory_failure_early_kill |
39 | - memory_failure_recovery | |
1da177e4 | 40 | - min_free_kbytes |
0ff38490 | 41 | - min_slab_ratio |
db0fb184 PM |
42 | - min_unmapped_ratio |
43 | - mmap_min_addr | |
d5dbac87 NA |
44 | - nr_hugepages |
45 | - nr_overcommit_hugepages | |
db0fb184 PM |
46 | - nr_trim_pages (only if CONFIG_MMU=n) |
47 | - numa_zonelist_order | |
48 | - oom_dump_tasks | |
49 | - oom_kill_allocating_task | |
49f0ce5f | 50 | - overcommit_kbytes |
db0fb184 PM |
51 | - overcommit_memory |
52 | - overcommit_ratio | |
53 | - page-cluster | |
54 | - panic_on_oom | |
55 | - percpu_pagelist_fraction | |
56 | - stat_interval | |
57 | - swappiness | |
c9b1d098 | 58 | - user_reserve_kbytes |
db0fb184 PM |
59 | - vfs_cache_pressure |
60 | - zone_reclaim_mode | |
61 | ||
1da177e4 LT |
62 | ============================================================== |
63 | ||
4eeab4f5 AS |
64 | admin_reserve_kbytes |
65 | ||
66 | The amount of free memory in the system that should be reserved for users | |
67 | with the capability cap_sys_admin. | |
68 | ||
69 | admin_reserve_kbytes defaults to min(3% of free pages, 8MB) | |
70 | ||
71 | That should provide enough for the admin to log in and kill a process, | |
72 | if necessary, under the default overcommit 'guess' mode. | |
73 | ||
74 | Systems running under overcommit 'never' should increase this to account | |
75 | for the full Virtual Memory Size of programs used to recover. Otherwise, | |
76 | root may not be able to log in to recover the system. | |
77 | ||
78 | How do you calculate a minimum useful reserve? | |
79 | ||
80 | sshd or login + bash (or some other shell) + top (or ps, kill, etc.) | |
81 | ||
82 | For overcommit 'guess', we can sum resident set sizes (RSS). | |
83 | On x86_64 this is about 8MB. | |
84 | ||
85 | For overcommit 'never', we can take the max of their virtual sizes (VSZ) | |
86 | and add the sum of their RSS. | |
87 | On x86_64 this is about 128MB. | |
88 | ||
89 | Changing this takes effect whenever an application requests memory. | |
90 | ||
91 | ============================================================== | |
92 | ||
db0fb184 | 93 | block_dump |
1da177e4 | 94 | |
db0fb184 PM |
95 | block_dump enables block I/O debugging when set to a nonzero value. More |
96 | information on block I/O debugging is in Documentation/laptops/laptop-mode.txt. | |
1da177e4 LT |
97 | |
98 | ============================================================== | |
99 | ||
76ab0f53 MG |
100 | compact_memory |
101 | ||
102 | Available only when CONFIG_COMPACTION is set. When 1 is written to the file, | |
103 | all zones are compacted such that free memory is available in contiguous | |
104 | blocks where possible. This can be important for example in the allocation of | |
105 | huge pages although processes will also directly compact memory as required. | |
106 | ||
107 | ============================================================== | |
108 | ||
db0fb184 | 109 | dirty_background_bytes |
1da177e4 | 110 | |
6601fac8 AB |
111 | Contains the amount of dirty memory at which the background kernel |
112 | flusher threads will start writeback. | |
1da177e4 | 113 | |
abffc020 AR |
114 | Note: dirty_background_bytes is the counterpart of dirty_background_ratio. Only |
115 | one of them may be specified at a time. When one sysctl is written it is | |
116 | immediately taken into account to evaluate the dirty memory limits and the | |
117 | other appears as 0 when read. | |
1da177e4 | 118 | |
db0fb184 | 119 | ============================================================== |
1da177e4 | 120 | |
db0fb184 | 121 | dirty_background_ratio |
1da177e4 | 122 | |
715ea41e ZL |
123 | Contains, as a percentage of total available memory that contains free pages |
124 | and reclaimable pages, the number of pages at which the background kernel | |
125 | flusher threads will start writing out dirty data. | |
126 | ||
127 | The total avaiable memory is not equal to total system memory. | |
1da177e4 | 128 | |
db0fb184 | 129 | ============================================================== |
1da177e4 | 130 | |
db0fb184 PM |
131 | dirty_bytes |
132 | ||
133 | Contains the amount of dirty memory at which a process generating disk writes | |
134 | will itself start writeback. | |
135 | ||
abffc020 AR |
136 | Note: dirty_bytes is the counterpart of dirty_ratio. Only one of them may be |
137 | specified at a time. When one sysctl is written it is immediately taken into | |
138 | account to evaluate the dirty memory limits and the other appears as 0 when | |
139 | read. | |
1da177e4 | 140 | |
9e4a5bda AR |
141 | Note: the minimum value allowed for dirty_bytes is two pages (in bytes); any |
142 | value lower than this limit will be ignored and the old configuration will be | |
143 | retained. | |
144 | ||
1da177e4 LT |
145 | ============================================================== |
146 | ||
db0fb184 | 147 | dirty_expire_centisecs |
1da177e4 | 148 | |
db0fb184 | 149 | This tunable is used to define when dirty data is old enough to be eligible |
6601fac8 AB |
150 | for writeout by the kernel flusher threads. It is expressed in 100'ths |
151 | of a second. Data which has been dirty in-memory for longer than this | |
152 | interval will be written out next time a flusher thread wakes up. | |
db0fb184 PM |
153 | |
154 | ============================================================== | |
155 | ||
156 | dirty_ratio | |
157 | ||
715ea41e ZL |
158 | Contains, as a percentage of total available memory that contains free pages |
159 | and reclaimable pages, the number of pages at which a process which is | |
160 | generating disk writes will itself start writing out dirty data. | |
161 | ||
162 | The total avaiable memory is not equal to total system memory. | |
1da177e4 LT |
163 | |
164 | ============================================================== | |
165 | ||
db0fb184 | 166 | dirty_writeback_centisecs |
1da177e4 | 167 | |
6601fac8 | 168 | The kernel flusher threads will periodically wake up and write `old' data |
db0fb184 PM |
169 | out to disk. This tunable expresses the interval between those wakeups, in |
170 | 100'ths of a second. | |
1da177e4 | 171 | |
db0fb184 | 172 | Setting this to zero disables periodic writeback altogether. |
1da177e4 LT |
173 | |
174 | ============================================================== | |
175 | ||
db0fb184 | 176 | drop_caches |
1da177e4 | 177 | |
db0fb184 PM |
178 | Writing to this will cause the kernel to drop clean caches, dentries and |
179 | inodes from memory, causing that memory to become free. | |
1da177e4 | 180 | |
db0fb184 PM |
181 | To free pagecache: |
182 | echo 1 > /proc/sys/vm/drop_caches | |
183 | To free dentries and inodes: | |
184 | echo 2 > /proc/sys/vm/drop_caches | |
185 | To free pagecache, dentries and inodes: | |
186 | echo 3 > /proc/sys/vm/drop_caches | |
1da177e4 | 187 | |
db0fb184 PM |
188 | As this is a non-destructive operation and dirty objects are not freeable, the |
189 | user should run `sync' first. | |
1da177e4 LT |
190 | |
191 | ============================================================== | |
192 | ||
5e771905 MG |
193 | extfrag_threshold |
194 | ||
195 | This parameter affects whether the kernel will compact memory or direct | |
196 | reclaim to satisfy a high-order allocation. /proc/extfrag_index shows what | |
197 | the fragmentation index for each order is in each zone in the system. Values | |
198 | tending towards 0 imply allocations would fail due to lack of memory, | |
199 | values towards 1000 imply failures are due to fragmentation and -1 implies | |
200 | that the allocation will succeed as long as watermarks are met. | |
201 | ||
202 | The kernel will not compact memory in a zone if the | |
203 | fragmentation index is <= extfrag_threshold. The default value is 500. | |
204 | ||
205 | ============================================================== | |
206 | ||
db0fb184 | 207 | hugepages_treat_as_movable |
1da177e4 | 208 | |
86cdb465 NH |
209 | This parameter controls whether we can allocate hugepages from ZONE_MOVABLE |
210 | or not. If set to non-zero, hugepages can be allocated from ZONE_MOVABLE. | |
211 | ZONE_MOVABLE is created when kernel boot parameter kernelcore= is specified, | |
212 | so this parameter has no effect if used without kernelcore=. | |
213 | ||
214 | Hugepage migration is now available in some situations which depend on the | |
215 | architecture and/or the hugepage size. If a hugepage supports migration, | |
216 | allocation from ZONE_MOVABLE is always enabled for the hugepage regardless | |
217 | of the value of this parameter. | |
218 | IOW, this parameter affects only non-migratable hugepages. | |
219 | ||
220 | Assuming that hugepages are not migratable in your system, one usecase of | |
221 | this parameter is that users can make hugepage pool more extensible by | |
222 | enabling the allocation from ZONE_MOVABLE. This is because on ZONE_MOVABLE | |
223 | page reclaim/migration/compaction work more and you can get contiguous | |
224 | memory more likely. Note that using ZONE_MOVABLE for non-migratable | |
225 | hugepages can do harm to other features like memory hotremove (because | |
226 | memory hotremove expects that memory blocks on ZONE_MOVABLE are always | |
227 | removable,) so it's a trade-off responsible for the users. | |
24950898 | 228 | |
8ad4b1fb RS |
229 | ============================================================== |
230 | ||
db0fb184 | 231 | hugetlb_shm_group |
8ad4b1fb | 232 | |
db0fb184 PM |
233 | hugetlb_shm_group contains group id that is allowed to create SysV |
234 | shared memory segment using hugetlb page. | |
8ad4b1fb | 235 | |
db0fb184 | 236 | ============================================================== |
8ad4b1fb | 237 | |
db0fb184 | 238 | laptop_mode |
1743660b | 239 | |
db0fb184 PM |
240 | laptop_mode is a knob that controls "laptop mode". All the things that are |
241 | controlled by this knob are discussed in Documentation/laptops/laptop-mode.txt. | |
1743660b | 242 | |
db0fb184 | 243 | ============================================================== |
1743660b | 244 | |
db0fb184 | 245 | legacy_va_layout |
1b2ffb78 | 246 | |
2174efb6 | 247 | If non-zero, this sysctl disables the new 32-bit mmap layout - the kernel |
db0fb184 | 248 | will use the legacy (2.4) layout for all processes. |
1b2ffb78 | 249 | |
db0fb184 | 250 | ============================================================== |
1b2ffb78 | 251 | |
db0fb184 PM |
252 | lowmem_reserve_ratio |
253 | ||
254 | For some specialised workloads on highmem machines it is dangerous for | |
255 | the kernel to allow process memory to be allocated from the "lowmem" | |
256 | zone. This is because that memory could then be pinned via the mlock() | |
257 | system call, or by unavailability of swapspace. | |
258 | ||
259 | And on large highmem machines this lack of reclaimable lowmem memory | |
260 | can be fatal. | |
261 | ||
262 | So the Linux page allocator has a mechanism which prevents allocations | |
263 | which _could_ use highmem from using too much lowmem. This means that | |
264 | a certain amount of lowmem is defended from the possibility of being | |
265 | captured into pinned user memory. | |
266 | ||
267 | (The same argument applies to the old 16 megabyte ISA DMA region. This | |
268 | mechanism will also defend that region from allocations which could use | |
269 | highmem or lowmem). | |
270 | ||
271 | The `lowmem_reserve_ratio' tunable determines how aggressive the kernel is | |
272 | in defending these lower zones. | |
273 | ||
274 | If you have a machine which uses highmem or ISA DMA and your | |
275 | applications are using mlock(), or if you are running with no swap then | |
276 | you probably should change the lowmem_reserve_ratio setting. | |
277 | ||
278 | The lowmem_reserve_ratio is an array. You can see them by reading this file. | |
279 | - | |
280 | % cat /proc/sys/vm/lowmem_reserve_ratio | |
281 | 256 256 32 | |
282 | - | |
283 | Note: # of this elements is one fewer than number of zones. Because the highest | |
284 | zone's value is not necessary for following calculation. | |
285 | ||
286 | But, these values are not used directly. The kernel calculates # of protection | |
287 | pages for each zones from them. These are shown as array of protection pages | |
288 | in /proc/zoneinfo like followings. (This is an example of x86-64 box). | |
289 | Each zone has an array of protection pages like this. | |
290 | ||
291 | - | |
292 | Node 0, zone DMA | |
293 | pages free 1355 | |
294 | min 3 | |
295 | low 3 | |
296 | high 4 | |
297 | : | |
298 | : | |
299 | numa_other 0 | |
300 | protection: (0, 2004, 2004, 2004) | |
301 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ | |
302 | pagesets | |
303 | cpu: 0 pcp: 0 | |
304 | : | |
305 | - | |
306 | These protections are added to score to judge whether this zone should be used | |
307 | for page allocation or should be reclaimed. | |
308 | ||
309 | In this example, if normal pages (index=2) are required to this DMA zone and | |
41858966 MG |
310 | watermark[WMARK_HIGH] is used for watermark, the kernel judges this zone should |
311 | not be used because pages_free(1355) is smaller than watermark + protection[2] | |
db0fb184 PM |
312 | (4 + 2004 = 2008). If this protection value is 0, this zone would be used for |
313 | normal page requirement. If requirement is DMA zone(index=0), protection[0] | |
314 | (=0) is used. | |
315 | ||
316 | zone[i]'s protection[j] is calculated by following expression. | |
317 | ||
318 | (i < j): | |
319 | zone[i]->protection[j] | |
320 | = (total sums of present_pages from zone[i+1] to zone[j] on the node) | |
321 | / lowmem_reserve_ratio[i]; | |
322 | (i = j): | |
323 | (should not be protected. = 0; | |
324 | (i > j): | |
325 | (not necessary, but looks 0) | |
326 | ||
327 | The default values of lowmem_reserve_ratio[i] are | |
328 | 256 (if zone[i] means DMA or DMA32 zone) | |
329 | 32 (others). | |
330 | As above expression, they are reciprocal number of ratio. | |
331 | 256 means 1/256. # of protection pages becomes about "0.39%" of total present | |
332 | pages of higher zones on the node. | |
333 | ||
334 | If you would like to protect more pages, smaller values are effective. | |
335 | The minimum value is 1 (1/1 -> 100%). | |
1b2ffb78 | 336 | |
db0fb184 | 337 | ============================================================== |
1b2ffb78 | 338 | |
db0fb184 | 339 | max_map_count: |
1743660b | 340 | |
db0fb184 PM |
341 | This file contains the maximum number of memory map areas a process |
342 | may have. Memory map areas are used as a side-effect of calling | |
343 | malloc, directly by mmap and mprotect, and also when loading shared | |
344 | libraries. | |
1743660b | 345 | |
db0fb184 PM |
346 | While most applications need less than a thousand maps, certain |
347 | programs, particularly malloc debuggers, may consume lots of them, | |
348 | e.g., up to one or two maps per allocation. | |
fadd8fbd | 349 | |
db0fb184 | 350 | The default value is 65536. |
9614634f | 351 | |
6a46079c AK |
352 | ============================================================= |
353 | ||
354 | memory_failure_early_kill: | |
355 | ||
356 | Control how to kill processes when uncorrected memory error (typically | |
357 | a 2bit error in a memory module) is detected in the background by hardware | |
358 | that cannot be handled by the kernel. In some cases (like the page | |
359 | still having a valid copy on disk) the kernel will handle the failure | |
360 | transparently without affecting any applications. But if there is | |
361 | no other uptodate copy of the data it will kill to prevent any data | |
362 | corruptions from propagating. | |
363 | ||
364 | 1: Kill all processes that have the corrupted and not reloadable page mapped | |
365 | as soon as the corruption is detected. Note this is not supported | |
366 | for a few types of pages, like kernel internally allocated data or | |
367 | the swap cache, but works for the majority of user pages. | |
368 | ||
369 | 0: Only unmap the corrupted page from all processes and only kill a process | |
370 | who tries to access it. | |
371 | ||
372 | The kill is done using a catchable SIGBUS with BUS_MCEERR_AO, so processes can | |
373 | handle this if they want to. | |
374 | ||
375 | This is only active on architectures/platforms with advanced machine | |
376 | check handling and depends on the hardware capabilities. | |
377 | ||
378 | Applications can override this setting individually with the PR_MCE_KILL prctl | |
379 | ||
380 | ============================================================== | |
381 | ||
382 | memory_failure_recovery | |
383 | ||
384 | Enable memory failure recovery (when supported by the platform) | |
385 | ||
386 | 1: Attempt recovery. | |
387 | ||
388 | 0: Always panic on a memory failure. | |
389 | ||
db0fb184 | 390 | ============================================================== |
9614634f | 391 | |
db0fb184 | 392 | min_free_kbytes: |
9614634f | 393 | |
db0fb184 | 394 | This is used to force the Linux VM to keep a minimum number |
41858966 MG |
395 | of kilobytes free. The VM uses this number to compute a |
396 | watermark[WMARK_MIN] value for each lowmem zone in the system. | |
397 | Each lowmem zone gets a number of reserved free pages based | |
398 | proportionally on its size. | |
db0fb184 PM |
399 | |
400 | Some minimal amount of memory is needed to satisfy PF_MEMALLOC | |
401 | allocations; if you set this to lower than 1024KB, your system will | |
402 | become subtly broken, and prone to deadlock under high loads. | |
403 | ||
404 | Setting this too high will OOM your machine instantly. | |
9614634f CL |
405 | |
406 | ============================================================= | |
407 | ||
0ff38490 CL |
408 | min_slab_ratio: |
409 | ||
410 | This is available only on NUMA kernels. | |
411 | ||
412 | A percentage of the total pages in each zone. On Zone reclaim | |
413 | (fallback from the local zone occurs) slabs will be reclaimed if more | |
414 | than this percentage of pages in a zone are reclaimable slab pages. | |
415 | This insures that the slab growth stays under control even in NUMA | |
416 | systems that rarely perform global reclaim. | |
417 | ||
418 | The default is 5 percent. | |
419 | ||
420 | Note that slab reclaim is triggered in a per zone / node fashion. | |
421 | The process of reclaiming slab memory is currently not node specific | |
422 | and may not be fast. | |
423 | ||
424 | ============================================================= | |
425 | ||
db0fb184 | 426 | min_unmapped_ratio: |
fadd8fbd | 427 | |
db0fb184 | 428 | This is available only on NUMA kernels. |
fadd8fbd | 429 | |
90afa5de MG |
430 | This is a percentage of the total pages in each zone. Zone reclaim will |
431 | only occur if more than this percentage of pages are in a state that | |
432 | zone_reclaim_mode allows to be reclaimed. | |
433 | ||
434 | If zone_reclaim_mode has the value 4 OR'd, then the percentage is compared | |
435 | against all file-backed unmapped pages including swapcache pages and tmpfs | |
436 | files. Otherwise, only unmapped pages backed by normal files but not tmpfs | |
437 | files and similar are considered. | |
2b744c01 | 438 | |
db0fb184 | 439 | The default is 1 percent. |
fadd8fbd | 440 | |
db0fb184 | 441 | ============================================================== |
2b744c01 | 442 | |
db0fb184 | 443 | mmap_min_addr |
ed032189 | 444 | |
db0fb184 | 445 | This file indicates the amount of address space which a user process will |
af901ca1 | 446 | be restricted from mmapping. Since kernel null dereference bugs could |
db0fb184 PM |
447 | accidentally operate based on the information in the first couple of pages |
448 | of memory userspace processes should not be allowed to write to them. By | |
449 | default this value is set to 0 and no protections will be enforced by the | |
450 | security module. Setting this value to something like 64k will allow the | |
451 | vast majority of applications to work correctly and provide defense in depth | |
452 | against future potential kernel bugs. | |
fe071d7e | 453 | |
db0fb184 | 454 | ============================================================== |
fef1bdd6 | 455 | |
db0fb184 | 456 | nr_hugepages |
fef1bdd6 | 457 | |
db0fb184 | 458 | Change the minimum size of the hugepage pool. |
fef1bdd6 | 459 | |
db0fb184 | 460 | See Documentation/vm/hugetlbpage.txt |
fef1bdd6 | 461 | |
db0fb184 | 462 | ============================================================== |
fef1bdd6 | 463 | |
db0fb184 | 464 | nr_overcommit_hugepages |
fef1bdd6 | 465 | |
db0fb184 PM |
466 | Change the maximum size of the hugepage pool. The maximum is |
467 | nr_hugepages + nr_overcommit_hugepages. | |
fe071d7e | 468 | |
db0fb184 | 469 | See Documentation/vm/hugetlbpage.txt |
fe071d7e | 470 | |
db0fb184 | 471 | ============================================================== |
fe071d7e | 472 | |
db0fb184 | 473 | nr_trim_pages |
ed032189 | 474 | |
db0fb184 PM |
475 | This is available only on NOMMU kernels. |
476 | ||
477 | This value adjusts the excess page trimming behaviour of power-of-2 aligned | |
478 | NOMMU mmap allocations. | |
479 | ||
480 | A value of 0 disables trimming of allocations entirely, while a value of 1 | |
481 | trims excess pages aggressively. Any value >= 1 acts as the watermark where | |
482 | trimming of allocations is initiated. | |
483 | ||
484 | The default value is 1. | |
485 | ||
486 | See Documentation/nommu-mmap.txt for more information. | |
ed032189 | 487 | |
f0c0b2b8 KH |
488 | ============================================================== |
489 | ||
490 | numa_zonelist_order | |
491 | ||
492 | This sysctl is only for NUMA. | |
493 | 'where the memory is allocated from' is controlled by zonelists. | |
494 | (This documentation ignores ZONE_HIGHMEM/ZONE_DMA32 for simple explanation. | |
495 | you may be able to read ZONE_DMA as ZONE_DMA32...) | |
496 | ||
497 | In non-NUMA case, a zonelist for GFP_KERNEL is ordered as following. | |
498 | ZONE_NORMAL -> ZONE_DMA | |
499 | This means that a memory allocation request for GFP_KERNEL will | |
500 | get memory from ZONE_DMA only when ZONE_NORMAL is not available. | |
501 | ||
502 | In NUMA case, you can think of following 2 types of order. | |
503 | Assume 2 node NUMA and below is zonelist of Node(0)'s GFP_KERNEL | |
504 | ||
505 | (A) Node(0) ZONE_NORMAL -> Node(0) ZONE_DMA -> Node(1) ZONE_NORMAL | |
506 | (B) Node(0) ZONE_NORMAL -> Node(1) ZONE_NORMAL -> Node(0) ZONE_DMA. | |
507 | ||
508 | Type(A) offers the best locality for processes on Node(0), but ZONE_DMA | |
509 | will be used before ZONE_NORMAL exhaustion. This increases possibility of | |
510 | out-of-memory(OOM) of ZONE_DMA because ZONE_DMA is tend to be small. | |
511 | ||
512 | Type(B) cannot offer the best locality but is more robust against OOM of | |
513 | the DMA zone. | |
514 | ||
515 | Type(A) is called as "Node" order. Type (B) is "Zone" order. | |
516 | ||
517 | "Node order" orders the zonelists by node, then by zone within each node. | |
5a3016a6 | 518 | Specify "[Nn]ode" for node order |
f0c0b2b8 KH |
519 | |
520 | "Zone Order" orders the zonelists by zone type, then by node within each | |
5a3016a6 | 521 | zone. Specify "[Zz]one" for zone order. |
f0c0b2b8 KH |
522 | |
523 | Specify "[Dd]efault" to request automatic configuration. Autoconfiguration | |
524 | will select "node" order in following case. | |
525 | (1) if the DMA zone does not exist or | |
526 | (2) if the DMA zone comprises greater than 50% of the available memory or | |
f8f191f1 | 527 | (3) if any node's DMA zone comprises greater than 70% of its local memory and |
f0c0b2b8 KH |
528 | the amount of local memory is big enough. |
529 | ||
530 | Otherwise, "zone" order will be selected. Default order is recommended unless | |
531 | this is causing problems for your system/application. | |
d5dbac87 NA |
532 | |
533 | ============================================================== | |
534 | ||
db0fb184 | 535 | oom_dump_tasks |
d5dbac87 | 536 | |
db0fb184 PM |
537 | Enables a system-wide task dump (excluding kernel threads) to be |
538 | produced when the kernel performs an OOM-killing and includes such | |
de34d965 DR |
539 | information as pid, uid, tgid, vm size, rss, nr_ptes, swapents, |
540 | oom_score_adj score, and name. This is helpful to determine why the | |
541 | OOM killer was invoked, to identify the rogue task that caused it, | |
542 | and to determine why the OOM killer chose the task it did to kill. | |
d5dbac87 | 543 | |
db0fb184 PM |
544 | If this is set to zero, this information is suppressed. On very |
545 | large systems with thousands of tasks it may not be feasible to dump | |
546 | the memory state information for each one. Such systems should not | |
547 | be forced to incur a performance penalty in OOM conditions when the | |
548 | information may not be desired. | |
549 | ||
550 | If this is set to non-zero, this information is shown whenever the | |
551 | OOM killer actually kills a memory-hogging task. | |
552 | ||
ad915c43 | 553 | The default value is 1 (enabled). |
d5dbac87 NA |
554 | |
555 | ============================================================== | |
556 | ||
db0fb184 | 557 | oom_kill_allocating_task |
d5dbac87 | 558 | |
db0fb184 PM |
559 | This enables or disables killing the OOM-triggering task in |
560 | out-of-memory situations. | |
d5dbac87 | 561 | |
db0fb184 PM |
562 | If this is set to zero, the OOM killer will scan through the entire |
563 | tasklist and select a task based on heuristics to kill. This normally | |
564 | selects a rogue memory-hogging task that frees up a large amount of | |
565 | memory when killed. | |
566 | ||
567 | If this is set to non-zero, the OOM killer simply kills the task that | |
568 | triggered the out-of-memory condition. This avoids the expensive | |
569 | tasklist scan. | |
570 | ||
571 | If panic_on_oom is selected, it takes precedence over whatever value | |
572 | is used in oom_kill_allocating_task. | |
573 | ||
574 | The default value is 0. | |
dd8632a1 PM |
575 | |
576 | ============================================================== | |
577 | ||
49f0ce5f JM |
578 | overcommit_kbytes: |
579 | ||
580 | When overcommit_memory is set to 2, the committed address space is not | |
581 | permitted to exceed swap plus this amount of physical RAM. See below. | |
582 | ||
583 | Note: overcommit_kbytes is the counterpart of overcommit_ratio. Only one | |
584 | of them may be specified at a time. Setting one disables the other (which | |
585 | then appears as 0 when read). | |
586 | ||
587 | ============================================================== | |
588 | ||
db0fb184 | 589 | overcommit_memory: |
dd8632a1 | 590 | |
db0fb184 | 591 | This value contains a flag that enables memory overcommitment. |
dd8632a1 | 592 | |
db0fb184 PM |
593 | When this flag is 0, the kernel attempts to estimate the amount |
594 | of free memory left when userspace requests more memory. | |
dd8632a1 | 595 | |
db0fb184 PM |
596 | When this flag is 1, the kernel pretends there is always enough |
597 | memory until it actually runs out. | |
dd8632a1 | 598 | |
db0fb184 PM |
599 | When this flag is 2, the kernel uses a "never overcommit" |
600 | policy that attempts to prevent any overcommit of memory. | |
c9b1d098 | 601 | Note that user_reserve_kbytes affects this policy. |
dd8632a1 | 602 | |
db0fb184 PM |
603 | This feature can be very useful because there are a lot of |
604 | programs that malloc() huge amounts of memory "just-in-case" | |
605 | and don't use much of it. | |
606 | ||
607 | The default value is 0. | |
608 | ||
609 | See Documentation/vm/overcommit-accounting and | |
610 | security/commoncap.c::cap_vm_enough_memory() for more information. | |
611 | ||
612 | ============================================================== | |
613 | ||
614 | overcommit_ratio: | |
615 | ||
616 | When overcommit_memory is set to 2, the committed address | |
617 | space is not permitted to exceed swap plus this percentage | |
618 | of physical RAM. See above. | |
619 | ||
620 | ============================================================== | |
621 | ||
622 | page-cluster | |
623 | ||
df858fa8 CE |
624 | page-cluster controls the number of pages up to which consecutive pages |
625 | are read in from swap in a single attempt. This is the swap counterpart | |
626 | to page cache readahead. | |
627 | The mentioned consecutivity is not in terms of virtual/physical addresses, | |
628 | but consecutive on swap space - that means they were swapped out together. | |
db0fb184 PM |
629 | |
630 | It is a logarithmic value - setting it to zero means "1 page", setting | |
631 | it to 1 means "2 pages", setting it to 2 means "4 pages", etc. | |
df858fa8 | 632 | Zero disables swap readahead completely. |
db0fb184 PM |
633 | |
634 | The default value is three (eight pages at a time). There may be some | |
635 | small benefits in tuning this to a different value if your workload is | |
636 | swap-intensive. | |
637 | ||
df858fa8 CE |
638 | Lower values mean lower latencies for initial faults, but at the same time |
639 | extra faults and I/O delays for following faults if they would have been part of | |
640 | that consecutive pages readahead would have brought in. | |
641 | ||
db0fb184 PM |
642 | ============================================================= |
643 | ||
644 | panic_on_oom | |
645 | ||
646 | This enables or disables panic on out-of-memory feature. | |
647 | ||
648 | If this is set to 0, the kernel will kill some rogue process, | |
649 | called oom_killer. Usually, oom_killer can kill rogue processes and | |
650 | system will survive. | |
651 | ||
652 | If this is set to 1, the kernel panics when out-of-memory happens. | |
653 | However, if a process limits using nodes by mempolicy/cpusets, | |
654 | and those nodes become memory exhaustion status, one process | |
655 | may be killed by oom-killer. No panic occurs in this case. | |
656 | Because other nodes' memory may be free. This means system total status | |
657 | may be not fatal yet. | |
658 | ||
659 | If this is set to 2, the kernel panics compulsorily even on the | |
daaf1e68 KH |
660 | above-mentioned. Even oom happens under memory cgroup, the whole |
661 | system panics. | |
db0fb184 PM |
662 | |
663 | The default value is 0. | |
664 | 1 and 2 are for failover of clustering. Please select either | |
665 | according to your policy of failover. | |
daaf1e68 KH |
666 | panic_on_oom=2+kdump gives you very strong tool to investigate |
667 | why oom happens. You can get snapshot. | |
db0fb184 PM |
668 | |
669 | ============================================================= | |
670 | ||
671 | percpu_pagelist_fraction | |
672 | ||
673 | This is the fraction of pages at most (high mark pcp->high) in each zone that | |
674 | are allocated for each per cpu page list. The min value for this is 8. It | |
675 | means that we don't allow more than 1/8th of pages in each zone to be | |
676 | allocated in any single per_cpu_pagelist. This entry only changes the value | |
677 | of hot per cpu pagelists. User can specify a number like 100 to allocate | |
678 | 1/100th of each zone to each per cpu page list. | |
679 | ||
680 | The batch value of each per cpu pagelist is also updated as a result. It is | |
681 | set to pcp->high/4. The upper limit of batch is (PAGE_SHIFT * 8) | |
682 | ||
683 | The initial value is zero. Kernel does not use this value at boot time to set | |
684 | the high water marks for each per cpu page list. | |
685 | ||
686 | ============================================================== | |
687 | ||
688 | stat_interval | |
689 | ||
690 | The time interval between which vm statistics are updated. The default | |
691 | is 1 second. | |
692 | ||
693 | ============================================================== | |
694 | ||
695 | swappiness | |
696 | ||
697 | This control is used to define how aggressive the kernel will swap | |
698 | memory pages. Higher values will increase agressiveness, lower values | |
8582cb96 AT |
699 | decrease the amount of swap. A value of 0 instructs the kernel not to |
700 | initiate swap until the amount of free and file-backed pages is less | |
701 | than the high water mark in a zone. | |
db0fb184 PM |
702 | |
703 | The default value is 60. | |
704 | ||
705 | ============================================================== | |
706 | ||
c9b1d098 AS |
707 | - user_reserve_kbytes |
708 | ||
709 | When overcommit_memory is set to 2, "never overommit" mode, reserve | |
710 | min(3% of current process size, user_reserve_kbytes) of free memory. | |
711 | This is intended to prevent a user from starting a single memory hogging | |
712 | process, such that they cannot recover (kill the hog). | |
713 | ||
714 | user_reserve_kbytes defaults to min(3% of the current process size, 128MB). | |
715 | ||
716 | If this is reduced to zero, then the user will be allowed to allocate | |
717 | all free memory with a single process, minus admin_reserve_kbytes. | |
718 | Any subsequent attempts to execute a command will result in | |
719 | "fork: Cannot allocate memory". | |
720 | ||
721 | Changing this takes effect whenever an application requests memory. | |
722 | ||
723 | ============================================================== | |
724 | ||
db0fb184 PM |
725 | vfs_cache_pressure |
726 | ------------------ | |
727 | ||
728 | Controls the tendency of the kernel to reclaim the memory which is used for | |
729 | caching of directory and inode objects. | |
730 | ||
731 | At the default value of vfs_cache_pressure=100 the kernel will attempt to | |
732 | reclaim dentries and inodes at a "fair" rate with respect to pagecache and | |
733 | swapcache reclaim. Decreasing vfs_cache_pressure causes the kernel to prefer | |
55c37a84 JK |
734 | to retain dentry and inode caches. When vfs_cache_pressure=0, the kernel will |
735 | never reclaim dentries and inodes due to memory pressure and this can easily | |
736 | lead to out-of-memory conditions. Increasing vfs_cache_pressure beyond 100 | |
db0fb184 PM |
737 | causes the kernel to prefer to reclaim dentries and inodes. |
738 | ||
739 | ============================================================== | |
740 | ||
741 | zone_reclaim_mode: | |
742 | ||
743 | Zone_reclaim_mode allows someone to set more or less aggressive approaches to | |
744 | reclaim memory when a zone runs out of memory. If it is set to zero then no | |
745 | zone reclaim occurs. Allocations will be satisfied from other zones / nodes | |
746 | in the system. | |
747 | ||
748 | This is value ORed together of | |
749 | ||
750 | 1 = Zone reclaim on | |
751 | 2 = Zone reclaim writes dirty pages out | |
752 | 4 = Zone reclaim swaps pages | |
753 | ||
754 | zone_reclaim_mode is set during bootup to 1 if it is determined that pages | |
755 | from remote zones will cause a measurable performance reduction. The | |
756 | page allocator will then reclaim easily reusable pages (those page | |
757 | cache pages that are currently not used) before allocating off node pages. | |
758 | ||
759 | It may be beneficial to switch off zone reclaim if the system is | |
760 | used for a file server and all of memory should be used for caching files | |
761 | from disk. In that case the caching effect is more important than | |
762 | data locality. | |
763 | ||
764 | Allowing zone reclaim to write out pages stops processes that are | |
765 | writing large amounts of data from dirtying pages on other nodes. Zone | |
766 | reclaim will write out dirty pages if a zone fills up and so effectively | |
767 | throttle the process. This may decrease the performance of a single process | |
768 | since it cannot use all of system memory to buffer the outgoing writes | |
769 | anymore but it preserve the memory on other nodes so that the performance | |
770 | of other processes running on other nodes will not be affected. | |
771 | ||
772 | Allowing regular swap effectively restricts allocations to the local | |
773 | node unless explicitly overridden by memory policies or cpuset | |
774 | configurations. | |
775 | ||
776 | ============ End of Document ================================= |