Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | Documentation for /proc/sys/vm/* kernel version 2.2.10 |
| 2 | (c) 1998, 1999, Rik van Riel <riel@nl.linux.org> |
| 3 | |
| 4 | For general info and legal blurb, please look in README. |
| 5 | |
| 6 | ============================================================== |
| 7 | |
| 8 | This file contains the documentation for the sysctl files in |
| 9 | /proc/sys/vm and is valid for Linux kernel version 2.2. |
| 10 | |
| 11 | The files in this directory can be used to tune the operation |
| 12 | of the virtual memory (VM) subsystem of the Linux kernel and |
| 13 | the writeout of dirty data to disk. |
| 14 | |
| 15 | Default values and initialization routines for most of these |
| 16 | files can be found in mm/swap.c. |
| 17 | |
| 18 | Currently, 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 Morton | 9d0243b | 2006-01-08 01:00:39 -0800 | [diff] [blame] | 29 | - drop-caches |
Christoph Lameter | 1743660 | 2006-01-18 17:42:32 -0800 | [diff] [blame] | 30 | - zone_reclaim_mode |
Christoph Lameter | 2a11ff0 | 2006-02-01 03:05:33 -0800 | [diff] [blame] | 31 | - zone_reclaim_interval |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 32 | |
| 33 | ============================================================== |
| 34 | |
| 35 | dirty_ratio, dirty_background_ratio, dirty_expire_centisecs, |
| 36 | dirty_writeback_centisecs, vfs_cache_pressure, laptop_mode, |
Andrew Morton | 9d0243b | 2006-01-08 01:00:39 -0800 | [diff] [blame] | 37 | block_dump, swap_token_timeout, drop-caches: |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 38 | |
| 39 | See Documentation/filesystems/proc.txt |
| 40 | |
| 41 | ============================================================== |
| 42 | |
| 43 | overcommit_memory: |
| 44 | |
| 45 | This value contains a flag that enables memory overcommitment. |
| 46 | |
| 47 | When this flag is 0, the kernel attempts to estimate the amount |
| 48 | of free memory left when userspace requests more memory. |
| 49 | |
| 50 | When this flag is 1, the kernel pretends there is always enough |
| 51 | memory until it actually runs out. |
| 52 | |
| 53 | When this flag is 2, the kernel uses a "never overcommit" |
| 54 | policy that attempts to prevent any overcommit of memory. |
| 55 | |
| 56 | This feature can be very useful because there are a lot of |
| 57 | programs that malloc() huge amounts of memory "just-in-case" |
| 58 | and don't use much of it. |
| 59 | |
| 60 | The default value is 0. |
| 61 | |
| 62 | See Documentation/vm/overcommit-accounting and |
| 63 | security/commoncap.c::cap_vm_enough_memory() for more information. |
| 64 | |
| 65 | ============================================================== |
| 66 | |
| 67 | overcommit_ratio: |
| 68 | |
| 69 | When overcommit_memory is set to 2, the committed address |
| 70 | space is not permitted to exceed swap plus this percentage |
| 71 | of physical RAM. See above. |
| 72 | |
| 73 | ============================================================== |
| 74 | |
| 75 | page-cluster: |
| 76 | |
| 77 | The Linux VM subsystem avoids excessive disk seeks by reading |
| 78 | multiple pages on a page fault. The number of pages it reads |
| 79 | is dependent on the amount of memory in your machine. |
| 80 | |
| 81 | The number of pages the kernel reads in at once is equal to |
| 82 | 2 ^ page-cluster. Values above 2 ^ 5 don't make much sense |
| 83 | for swap because we only cluster swap data in 32-page groups. |
| 84 | |
| 85 | ============================================================== |
| 86 | |
| 87 | max_map_count: |
| 88 | |
| 89 | This file contains the maximum number of memory map areas a process |
| 90 | may have. Memory map areas are used as a side-effect of calling |
| 91 | malloc, directly by mmap and mprotect, and also when loading shared |
| 92 | libraries. |
| 93 | |
| 94 | While most applications need less than a thousand maps, certain |
| 95 | programs, particularly malloc debuggers, may consume lots of them, |
| 96 | e.g., up to one or two maps per allocation. |
| 97 | |
| 98 | The default value is 65536. |
| 99 | |
| 100 | ============================================================== |
| 101 | |
| 102 | min_free_kbytes: |
| 103 | |
| 104 | This is used to force the Linux VM to keep a minimum number |
| 105 | of kilobytes free. The VM uses this number to compute a pages_min |
| 106 | value for each lowmem zone in the system. Each lowmem zone gets |
| 107 | a number of reserved free pages based proportionally on its size. |
Rohit Seth | 8ad4b1f | 2006-01-08 01:00:40 -0800 | [diff] [blame] | 108 | |
| 109 | ============================================================== |
| 110 | |
| 111 | percpu_pagelist_fraction |
| 112 | |
| 113 | This is the fraction of pages at most (high mark pcp->high) in each zone that |
| 114 | are allocated for each per cpu page list. The min value for this is 8. It |
| 115 | means that we don't allow more than 1/8th of pages in each zone to be |
| 116 | allocated in any single per_cpu_pagelist. This entry only changes the value |
| 117 | of hot per cpu pagelists. User can specify a number like 100 to allocate |
| 118 | 1/100th of each zone to each per cpu page list. |
| 119 | |
| 120 | The batch value of each per cpu pagelist is also updated as a result. It is |
| 121 | set to pcp->high/4. The upper limit of batch is (PAGE_SHIFT * 8) |
| 122 | |
| 123 | The initial value is zero. Kernel does not use this value at boot time to set |
| 124 | the high water marks for each per cpu page list. |
Christoph Lameter | 1743660 | 2006-01-18 17:42:32 -0800 | [diff] [blame] | 125 | |
| 126 | =============================================================== |
| 127 | |
| 128 | zone_reclaim_mode: |
| 129 | |
Christoph Lameter | 1b2ffb7 | 2006-02-01 03:05:34 -0800 | [diff] [blame^] | 130 | Zone_reclaim_mode allows to set more or less agressive approaches to |
| 131 | reclaim memory when a zone runs out of memory. If it is set to zero then no |
| 132 | zone reclaim occurs. Allocations will be satisfied from other zones / nodes |
| 133 | in the system. |
| 134 | |
| 135 | This is value ORed together of |
| 136 | |
| 137 | 1 = Zone reclaim on |
| 138 | 2 = Zone reclaim writes dirty pages out |
| 139 | 4 = Zone reclaim swaps pages |
| 140 | |
| 141 | zone_reclaim_mode is set during bootup to 1 if it is determined that pages |
| 142 | from remote zones will cause a measurable performance reduction. The |
Christoph Lameter | 1743660 | 2006-01-18 17:42:32 -0800 | [diff] [blame] | 143 | page allocator will then reclaim easily reusable pages (those page |
Christoph Lameter | 1b2ffb7 | 2006-02-01 03:05:34 -0800 | [diff] [blame^] | 144 | cache pages that are currently not used) before allocating off node pages. |
Christoph Lameter | 1743660 | 2006-01-18 17:42:32 -0800 | [diff] [blame] | 145 | |
Christoph Lameter | 1b2ffb7 | 2006-02-01 03:05:34 -0800 | [diff] [blame^] | 146 | It may be beneficial to switch off zone reclaim if the system is |
| 147 | used for a file server and all of memory should be used for caching files |
| 148 | from disk. In that case the caching effect is more important than |
| 149 | data locality. |
Christoph Lameter | 1743660 | 2006-01-18 17:42:32 -0800 | [diff] [blame] | 150 | |
Christoph Lameter | 1b2ffb7 | 2006-02-01 03:05:34 -0800 | [diff] [blame^] | 151 | Allowing zone reclaim to write out pages stops processes that are |
| 152 | writing large amounts of data from dirtying pages on other nodes. Zone |
| 153 | reclaim will write out dirty pages if a zone fills up and so effectively |
| 154 | throttle the process. This may decrease the performance of a single process |
| 155 | since it cannot use all of system memory to buffer the outgoing writes |
| 156 | anymore but it preserve the memory on other nodes so that the performance |
| 157 | of other processes running on other nodes will not be affected. |
| 158 | |
| 159 | Allowing regular swap effectively restricts allocations to the local |
| 160 | node unless explicitly overridden by memory policies or cpuset |
| 161 | configurations. |
| 162 | |
Christoph Lameter | 2a11ff0 | 2006-02-01 03:05:33 -0800 | [diff] [blame] | 163 | ================================================================ |
| 164 | |
| 165 | zone_reclaim_interval: |
| 166 | |
| 167 | The time allowed for off node allocations after zone reclaim |
| 168 | has failed to reclaim enough pages to allow a local allocation. |
| 169 | |
| 170 | Time is set in seconds and set by default to 30 seconds. |
| 171 | |
| 172 | Reduce the interval if undesired off node allocations occur. However, too |
| 173 | frequent scans will have a negative impact onoff node allocation performance. |
Christoph Lameter | 1743660 | 2006-01-18 17:42:32 -0800 | [diff] [blame] | 174 | |