On 01/12/09 09:57 Jonathan Miles said the following:

> I've attached a good example which shows Xorg invoking the OOM killer
> when the page-cache is using ~1.3GB out of 2GB. My understanding from
> reading linux-mm.org is that the kernel actually prefers to swap than
> reclaim from the page-cache... is this accurate? However, what if there
> is no swap, or it's configured not to use it? Should the kernel then try
> to reclaim instead?

And here's another (attached), with a slightly different initial code
path, where hald has caused the OOM killer to be invoked even though
(again) there was ~1.3GB out of 2GB used by the page-cache.

[66568.252937] Call Trace:
[66568.252963]  [<c01b5a0f>] oom_kill_process+0x9f/0x250
[66568.252981]  [<c01b601e>] ? select_bad_process+0xbe/0xf0
[66568.252989]  [<c01b60a1>] __out_of_memory+0x51/0xa0
[66568.252997]  [<c01b6143>] out_of_memory+0x53/0xb0
[66568.253045]  [<c01b83fe>] __alloc_pages_slowpath+0x42e/0x480
[66568.253055]  [<c01b855f>] __alloc_pages_nodemask+0x10f/0x120
[66568.253077]  [<c01ca076>] do_anonymous_page+0x66/0x200
...

Hopefully I'll get some time to RTFS, but at the moment I'm trying to
understand whether or not this is *intended* behaviour. When there's no
swap available, should the kernel be calling the OOM killer instead of
trying to reclaim memory from buffers/cache?

Thanks,

-- 
Jonathan Miles <jon@cybus.co.uk>
http://www.linkedin.com/in/jonmiles