]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
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 | |
9d0243bc | 29 | - drop-caches |
1743660b | 30 | - zone_reclaim_mode |
9614634f | 31 | - min_unmapped_ratio |
0ff38490 | 32 | - min_slab_ratio |
fadd8fbd | 33 | - panic_on_oom |
ed032189 | 34 | - mmap_min_address |
1da177e4 LT |
35 | |
36 | ============================================================== | |
37 | ||
38 | dirty_ratio, dirty_background_ratio, dirty_expire_centisecs, | |
39 | dirty_writeback_centisecs, vfs_cache_pressure, laptop_mode, | |
9d0243bc | 40 | block_dump, swap_token_timeout, drop-caches: |
1da177e4 LT |
41 | |
42 | See Documentation/filesystems/proc.txt | |
43 | ||
44 | ============================================================== | |
45 | ||
46 | overcommit_memory: | |
47 | ||
48 | This value contains a flag that enables memory overcommitment. | |
49 | ||
50 | When this flag is 0, the kernel attempts to estimate the amount | |
51 | of free memory left when userspace requests more memory. | |
52 | ||
53 | When this flag is 1, the kernel pretends there is always enough | |
54 | memory until it actually runs out. | |
55 | ||
56 | When this flag is 2, the kernel uses a "never overcommit" | |
57 | policy that attempts to prevent any overcommit of memory. | |
58 | ||
59 | This feature can be very useful because there are a lot of | |
60 | programs that malloc() huge amounts of memory "just-in-case" | |
61 | and don't use much of it. | |
62 | ||
63 | The default value is 0. | |
64 | ||
65 | See Documentation/vm/overcommit-accounting and | |
66 | security/commoncap.c::cap_vm_enough_memory() for more information. | |
67 | ||
68 | ============================================================== | |
69 | ||
70 | overcommit_ratio: | |
71 | ||
72 | When overcommit_memory is set to 2, the committed address | |
73 | space is not permitted to exceed swap plus this percentage | |
74 | of physical RAM. See above. | |
75 | ||
76 | ============================================================== | |
77 | ||
78 | page-cluster: | |
79 | ||
80 | The Linux VM subsystem avoids excessive disk seeks by reading | |
81 | multiple pages on a page fault. The number of pages it reads | |
82 | is dependent on the amount of memory in your machine. | |
83 | ||
84 | The number of pages the kernel reads in at once is equal to | |
85 | 2 ^ page-cluster. Values above 2 ^ 5 don't make much sense | |
86 | for swap because we only cluster swap data in 32-page groups. | |
87 | ||
88 | ============================================================== | |
89 | ||
90 | max_map_count: | |
91 | ||
92 | This file contains the maximum number of memory map areas a process | |
93 | may have. Memory map areas are used as a side-effect of calling | |
94 | malloc, directly by mmap and mprotect, and also when loading shared | |
95 | libraries. | |
96 | ||
97 | While most applications need less than a thousand maps, certain | |
98 | programs, particularly malloc debuggers, may consume lots of them, | |
99 | e.g., up to one or two maps per allocation. | |
100 | ||
101 | The default value is 65536. | |
102 | ||
103 | ============================================================== | |
104 | ||
105 | min_free_kbytes: | |
106 | ||
107 | This is used to force the Linux VM to keep a minimum number | |
108 | of kilobytes free. The VM uses this number to compute a pages_min | |
109 | value for each lowmem zone in the system. Each lowmem zone gets | |
110 | a number of reserved free pages based proportionally on its size. | |
8ad4b1fb RS |
111 | |
112 | ============================================================== | |
113 | ||
114 | percpu_pagelist_fraction | |
115 | ||
116 | This is the fraction of pages at most (high mark pcp->high) in each zone that | |
117 | are allocated for each per cpu page list. The min value for this is 8. It | |
118 | means that we don't allow more than 1/8th of pages in each zone to be | |
119 | allocated in any single per_cpu_pagelist. This entry only changes the value | |
120 | of hot per cpu pagelists. User can specify a number like 100 to allocate | |
121 | 1/100th of each zone to each per cpu page list. | |
122 | ||
123 | The batch value of each per cpu pagelist is also updated as a result. It is | |
124 | set to pcp->high/4. The upper limit of batch is (PAGE_SHIFT * 8) | |
125 | ||
126 | The initial value is zero. Kernel does not use this value at boot time to set | |
127 | the high water marks for each per cpu page list. | |
1743660b CL |
128 | |
129 | =============================================================== | |
130 | ||
131 | zone_reclaim_mode: | |
132 | ||
5d3f083d | 133 | Zone_reclaim_mode allows someone to set more or less aggressive approaches to |
1b2ffb78 CL |
134 | reclaim memory when a zone runs out of memory. If it is set to zero then no |
135 | zone reclaim occurs. Allocations will be satisfied from other zones / nodes | |
136 | in the system. | |
137 | ||
138 | This is value ORed together of | |
139 | ||
140 | 1 = Zone reclaim on | |
141 | 2 = Zone reclaim writes dirty pages out | |
142 | 4 = Zone reclaim swaps pages | |
143 | ||
144 | zone_reclaim_mode is set during bootup to 1 if it is determined that pages | |
145 | from remote zones will cause a measurable performance reduction. The | |
1743660b | 146 | page allocator will then reclaim easily reusable pages (those page |
1b2ffb78 CL |
147 | cache pages that are currently not used) before allocating off node pages. |
148 | ||
149 | It may be beneficial to switch off zone reclaim if the system is | |
150 | used for a file server and all of memory should be used for caching files | |
151 | from disk. In that case the caching effect is more important than | |
152 | data locality. | |
153 | ||
154 | Allowing zone reclaim to write out pages stops processes that are | |
155 | writing large amounts of data from dirtying pages on other nodes. Zone | |
156 | reclaim will write out dirty pages if a zone fills up and so effectively | |
157 | throttle the process. This may decrease the performance of a single process | |
158 | since it cannot use all of system memory to buffer the outgoing writes | |
159 | anymore but it preserve the memory on other nodes so that the performance | |
160 | of other processes running on other nodes will not be affected. | |
1743660b | 161 | |
1b2ffb78 CL |
162 | Allowing regular swap effectively restricts allocations to the local |
163 | node unless explicitly overridden by memory policies or cpuset | |
164 | configurations. | |
1743660b | 165 | |
fadd8fbd KH |
166 | ============================================================= |
167 | ||
9614634f CL |
168 | min_unmapped_ratio: |
169 | ||
170 | This is available only on NUMA kernels. | |
171 | ||
0ff38490 | 172 | A percentage of the total pages in each zone. Zone reclaim will only |
9614634f CL |
173 | occur if more than this percentage of pages are file backed and unmapped. |
174 | This is to insure that a minimal amount of local pages is still available for | |
175 | file I/O even if the node is overallocated. | |
176 | ||
177 | The default is 1 percent. | |
178 | ||
179 | ============================================================= | |
180 | ||
0ff38490 CL |
181 | min_slab_ratio: |
182 | ||
183 | This is available only on NUMA kernels. | |
184 | ||
185 | A percentage of the total pages in each zone. On Zone reclaim | |
186 | (fallback from the local zone occurs) slabs will be reclaimed if more | |
187 | than this percentage of pages in a zone are reclaimable slab pages. | |
188 | This insures that the slab growth stays under control even in NUMA | |
189 | systems that rarely perform global reclaim. | |
190 | ||
191 | The default is 5 percent. | |
192 | ||
193 | Note that slab reclaim is triggered in a per zone / node fashion. | |
194 | The process of reclaiming slab memory is currently not node specific | |
195 | and may not be fast. | |
196 | ||
197 | ============================================================= | |
198 | ||
fadd8fbd KH |
199 | panic_on_oom |
200 | ||
2b744c01 | 201 | This enables or disables panic on out-of-memory feature. |
fadd8fbd | 202 | |
2b744c01 YG |
203 | If this is set to 0, the kernel will kill some rogue process, |
204 | called oom_killer. Usually, oom_killer can kill rogue processes and | |
205 | system will survive. | |
206 | ||
207 | If this is set to 1, the kernel panics when out-of-memory happens. | |
208 | However, if a process limits using nodes by mempolicy/cpusets, | |
209 | and those nodes become memory exhaustion status, one process | |
210 | may be killed by oom-killer. No panic occurs in this case. | |
211 | Because other nodes' memory may be free. This means system total status | |
212 | may be not fatal yet. | |
fadd8fbd | 213 | |
2b744c01 YG |
214 | If this is set to 2, the kernel panics compulsorily even on the |
215 | above-mentioned. | |
216 | ||
217 | The default value is 0. | |
218 | 1 and 2 are for failover of clustering. Please select either | |
219 | according to your policy of failover. | |
ed032189 EP |
220 | |
221 | ============================================================== | |
222 | ||
223 | mmap_min_addr | |
224 | ||
225 | This file indicates the amount of address space which a user process will | |
226 | be restricted from mmaping. Since kernel null dereference bugs could | |
227 | accidentally operate based on the information in the first couple of pages | |
228 | of memory userspace processes should not be allowed to write to them. By | |
229 | default this value is set to 0 and no protections will be enforced by the | |
230 | security module. Setting this value to something like 64k will allow the | |
231 | vast majority of applications to work correctly and provide defense in depth | |
232 | against future potential kernel bugs. | |
233 |