blob: 4bca2a3d9174677b3275d62c93d01b8d99217b1a [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001Documentation for /proc/sys/vm/* kernel version 2.2.10
2 (c) 1998, 1999, Rik van Riel <riel@nl.linux.org>
3
4For general info and legal blurb, please look in README.
5
6==============================================================
7
8This file contains the documentation for the sysctl files in
9/proc/sys/vm and is valid for Linux kernel version 2.2.
10
11The files in this directory can be used to tune the operation
12of the virtual memory (VM) subsystem of the Linux kernel and
13the writeout of dirty data to disk.
14
15Default values and initialization routines for most of these
16files can be found in mm/swap.c.
17
18Currently, these files are in /proc/sys/vm:
19- overcommit_memory
20- page-cluster
21- dirty_ratio
22- dirty_background_ratio
23- dirty_expire_centisecs
24- dirty_writeback_centisecs
25- max_map_count
26- min_free_kbytes
27- laptop_mode
28- block_dump
Andrew Morton9d0243b2006-01-08 01:00:39 -080029- drop-caches
Christoph Lameter17436602006-01-18 17:42:32 -080030- zone_reclaim_mode
Christoph Lameter2a11ff02006-02-01 03:05:33 -080031- zone_reclaim_interval
Linus Torvalds1da177e2005-04-16 15:20:36 -070032
33==============================================================
34
35dirty_ratio, dirty_background_ratio, dirty_expire_centisecs,
36dirty_writeback_centisecs, vfs_cache_pressure, laptop_mode,
Andrew Morton9d0243b2006-01-08 01:00:39 -080037block_dump, swap_token_timeout, drop-caches:
Linus Torvalds1da177e2005-04-16 15:20:36 -070038
39See Documentation/filesystems/proc.txt
40
41==============================================================
42
43overcommit_memory:
44
45This value contains a flag that enables memory overcommitment.
46
47When this flag is 0, the kernel attempts to estimate the amount
48of free memory left when userspace requests more memory.
49
50When this flag is 1, the kernel pretends there is always enough
51memory until it actually runs out.
52
53When this flag is 2, the kernel uses a "never overcommit"
54policy that attempts to prevent any overcommit of memory.
55
56This feature can be very useful because there are a lot of
57programs that malloc() huge amounts of memory "just-in-case"
58and don't use much of it.
59
60The default value is 0.
61
62See Documentation/vm/overcommit-accounting and
63security/commoncap.c::cap_vm_enough_memory() for more information.
64
65==============================================================
66
67overcommit_ratio:
68
69When overcommit_memory is set to 2, the committed address
70space is not permitted to exceed swap plus this percentage
71of physical RAM. See above.
72
73==============================================================
74
75page-cluster:
76
77The Linux VM subsystem avoids excessive disk seeks by reading
78multiple pages on a page fault. The number of pages it reads
79is dependent on the amount of memory in your machine.
80
81The number of pages the kernel reads in at once is equal to
822 ^ page-cluster. Values above 2 ^ 5 don't make much sense
83for swap because we only cluster swap data in 32-page groups.
84
85==============================================================
86
87max_map_count:
88
89This file contains the maximum number of memory map areas a process
90may have. Memory map areas are used as a side-effect of calling
91malloc, directly by mmap and mprotect, and also when loading shared
92libraries.
93
94While most applications need less than a thousand maps, certain
95programs, particularly malloc debuggers, may consume lots of them,
96e.g., up to one or two maps per allocation.
97
98The default value is 65536.
99
100==============================================================
101
102min_free_kbytes:
103
104This is used to force the Linux VM to keep a minimum number
105of kilobytes free. The VM uses this number to compute a pages_min
106value for each lowmem zone in the system. Each lowmem zone gets
107a number of reserved free pages based proportionally on its size.
Rohit Seth8ad4b1f2006-01-08 01:00:40 -0800108
109==============================================================
110
111percpu_pagelist_fraction
112
113This is the fraction of pages at most (high mark pcp->high) in each zone that
114are allocated for each per cpu page list. The min value for this is 8. It
115means that we don't allow more than 1/8th of pages in each zone to be
116allocated in any single per_cpu_pagelist. This entry only changes the value
117of hot per cpu pagelists. User can specify a number like 100 to allocate
1181/100th of each zone to each per cpu page list.
119
120The batch value of each per cpu pagelist is also updated as a result. It is
121set to pcp->high/4. The upper limit of batch is (PAGE_SHIFT * 8)
122
123The initial value is zero. Kernel does not use this value at boot time to set
124the high water marks for each per cpu page list.
Christoph Lameter17436602006-01-18 17:42:32 -0800125
126===============================================================
127
128zone_reclaim_mode:
129
Christoph Lameter1b2ffb72006-02-01 03:05:34 -0800130Zone_reclaim_mode allows to set more or less agressive approaches to
131reclaim memory when a zone runs out of memory. If it is set to zero then no
132zone reclaim occurs. Allocations will be satisfied from other zones / nodes
133in the system.
134
135This is value ORed together of
136
1371 = Zone reclaim on
1382 = Zone reclaim writes dirty pages out
1394 = Zone reclaim swaps pages
140
141zone_reclaim_mode is set during bootup to 1 if it is determined that pages
142from remote zones will cause a measurable performance reduction. The
Christoph Lameter17436602006-01-18 17:42:32 -0800143page allocator will then reclaim easily reusable pages (those page
Christoph Lameter1b2ffb72006-02-01 03:05:34 -0800144cache pages that are currently not used) before allocating off node pages.
Christoph Lameter17436602006-01-18 17:42:32 -0800145
Christoph Lameter1b2ffb72006-02-01 03:05:34 -0800146It may be beneficial to switch off zone reclaim if the system is
147used for a file server and all of memory should be used for caching files
148from disk. In that case the caching effect is more important than
149data locality.
Christoph Lameter17436602006-01-18 17:42:32 -0800150
Christoph Lameter1b2ffb72006-02-01 03:05:34 -0800151Allowing zone reclaim to write out pages stops processes that are
152writing large amounts of data from dirtying pages on other nodes. Zone
153reclaim will write out dirty pages if a zone fills up and so effectively
154throttle the process. This may decrease the performance of a single process
155since it cannot use all of system memory to buffer the outgoing writes
156anymore but it preserve the memory on other nodes so that the performance
157of other processes running on other nodes will not be affected.
158
159Allowing regular swap effectively restricts allocations to the local
160node unless explicitly overridden by memory policies or cpuset
161configurations.
162
Christoph Lameter2a11ff02006-02-01 03:05:33 -0800163================================================================
164
165zone_reclaim_interval:
166
167The time allowed for off node allocations after zone reclaim
168has failed to reclaim enough pages to allow a local allocation.
169
170Time is set in seconds and set by default to 30 seconds.
171
172Reduce the interval if undesired off node allocations occur. However, too
173frequent scans will have a negative impact onoff node allocation performance.
Christoph Lameter17436602006-01-18 17:42:32 -0800174