lkml.org 
[lkml]   [2009]   [Aug]   [31]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: oom-killer works,but the memory cannot be freed
On Fri, 28 Aug 2009, Bill Xie wrote:

> and the oom killer message attached. It seems when at this situation,
>
> echo 3 > /proc/sys/vm/drop_caches , cannot release the memory occupied
> by the application.
>
> hotplug invoked oom-killer: gfp_mask=0x1200d2, order=0, oomkilladj=0
> Call Trace:
> [<800207dc>] dump_stack+0x8/0x34
> [<8007cb30>] oom_kill_process+0x138/0x148
> [<8007d1f0>] out_of_memory+0x1f0/0x340
> [<800806a4>] __alloc_pages+0x2a4/0x3d0
> [<8008ebd8>] do_wp_page+0x12c/0x8ec
> [<800913a0>] handle_mm_fault+0x718/0x938
> [<80024a1c>] do_page_fault+0x12c/0x3b0
> [<800196e4>] ret_from_exception+0x0/0x10
>
> Mem-info:
> Normal per-cpu:
> CPU 0: Hot: hi: 18, btch: 3 usd: 3 Cold: hi: 6, btch: 1 usd: 0
> Active:7767 inactive:2992 dirty:0 writeback:0 unstable:0
> free:252 slab:676 mapped:95 pagetables:99 bounce:0
> Normal free:1008kB min:1016kB low:1268kB high:1524kB active:31068kB inactive:11968kB press
> lowmem_reserve[]: 0 0
> Normal: 0*4kB 2*8kB 2*16kB 0*32kB 1*64kB 1*128kB 1*256kB 1*512kB 0*1024kB 0*2048kB 0*4096B
> Swap cache: add 0, delete 0, find 0/0, race 0+0
> Free swap = 0kB
> Total swap = 0kB
> Free swap: 0kB
> 16384 pages of RAM
> 0 pages of HIGHMEM
> 2467 reserved pages
> 2376 pages shared
> 0 pages swap cached
> Out of memory (oom_kill_allocating_task): kill process 511 (hotplug) score 0 or a child

The key part of these diagnostic messages is the last line, where it says
you're using oom_kill_allocating_task. Notice the badness score of the
task chosen for oom kill, hotplug, is 0, meaning that it's not going to
lead to any future memory freeing.

We should probably check if badness(current) is 0 before killing it when
oom_kill_allocating_task is enabled to ensure something worthwhile is
actually killed.

In the meantime, try

echo 0 > /proc/sys/vm/oom_kill_allocating_task

[ In the future, please don't include attachments to messages to LKML
unless they are egregiously long, we can handle smaller sizes inline. ]


\
 
 \ /
  Last update: 2009-09-01 01:03    [W:0.041 / U:2.596 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site