overcommit_memory/overcommit_ratio/overcommit_kbytes

最新推荐文章于 2025-03-22 11:00:08 发布

adaptiver

最新推荐文章于 2025-03-22 11:00:08 发布

阅读量1.5k

点赞数

分类专栏： kernel android

android 同时被 2 个专栏收录

24 篇文章

订阅专栏

kernel

18 篇文章

订阅专栏

http://blog.youkuaiyun.com/jollyjumper/article/details/24127009

新配的服务器内存128G,但在空余内存还有20G,page cache 60G时jps启动不了，提示java heap space,c中调用malloc无法分配5G以上的内存。

发现是系统参数overcommit_memory和overcommit_ratio搞的鬼。

grep -i commit /proc/meminfo

看到CommitLimit和Committed_As参数。

CommitLimit是一个内存分配上限,CommitLimit= 物理内存 * overcommit_ratio(默认50，即50%) + swap大小

Committed_As是已经分配的内存大小。

overcommit_memory参数就是控制分配内存是否可以超过CommitLimit，默认是0,即启发式的overcommitting handle,会尽量减少swap的使用,root可以分配比一般用户略多的内存。1表示允许超过CommitLimit,2表示不允许超过CommitLimit。

Attention:

overcommit_memory is set to 1 for almostall android platform.

Following is from Doc of kernel

overcommit_kbytes:

When overcommit_memory is set to 2, thecommitted address space is not permitted to exceed swap plus this amount ofphysical RAM. See below.

Note: overcommit_kbytes is the counterpartof overcommit_ratio. Only one of them may be specified at a time. Setting onedisables the other (which then appears as 0 when read).

==============================================================

overcommit_memory:

This value contains a flag that enablesmemory overcommitment.

When this flag is 0, the kernel attempts toestimate the amount of free memory left when userspace requests more memory.

When this flag is 1, the kernel pretendsthere is always enough memory until it actually runs out.

When this flag is 2, the kernel uses a"never overcommit" policy that attempts to prevent any overcommit ofmemory. Note that user_reserve_kbytes affects this policy.

This feature can be very useful becausethere are a lot of programs that malloc() huge amounts of memory"just-in-case" and don't use much of it.

The default value is 0.

See Documentation/vm/overcommit-accountingand security/commoncap.c::cap_vm_enough_memory() for more information.

==============================================================

overcommit_ratio:

When overcommit_memory is set to 2, thecommitted address space is not permitted to exceed swap plus this percentage ofphysical RAM. See above.

The Linux kernel supports the followingovercommit handling modes

0 - Heuristic overcommithandling. Obvious overcommits of

address space are refused. Usedfor a typical system. It

ensures a seriously wildallocation fails while allowing

overcommit to reduce swapusage. root is allowed to

allocate slightly more memoryin this mode. This is the

default.

1 - Always overcommit.Appropriate for some scientific

applications. Classic exampleis code using sparse arrays

and just relying on the virtualmemory consisting almost

entirely of zero pages.

2 - Don't overcommit. Thetotal address space commit

for the system is not permittedto exceed swap + a

configurable amount (default is50%) of physical RAM.

Depending on the amount youuse, in most situations

this means a process will notbe killed while accessing

pages but will receive errorson memory allocation as

appropriate.

Useful for applications thatwant to guarantee their

memory allocations will be available in thefuture

without having to initializeevery page.

The current overcommit limit and amountcommitted are viewable in

/proc/meminfo as CommitLimit andCommitted_AS respectively.

The overcommit amount can be set via`vm.overcommit_ratio' (percentage)

or `vm.overcommit_kbytes' (absolute value).

overcommit_kbytes is the counterpart ofovercommit_ratio. Only one of them may be specified at a time. Setting onedisables the other (which then appears as 0 when read).

It can be seen from following opertations:

# cat overcommit_ratio

# cat overcommit_memory

# cat overcommit_kbytes

# echo 910221312 > overcommit_kbytes

# cat overcommit_kbytes

910221312

# cat overcommit_ratio

# echo 60 > overcommit_ratio

# cat overcommit_kbytes

# cat overcommit_ratio

shell@aaa:/proc/sys/vm # cat /proc/meminfo| grep Commit