--===============0227967578==
Content-Type: multipart/signed;
  boundary="nextPart15382529.vCPLkiFksl";
  protocol="application/pgp-signature";
  micalg=pgp-sha1
Content-Transfer-Encoding: 7bit

--nextPart15382529.vCPLkiFksl
Content-Type: Text/Plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

On Saturday 21 November 2009 17:38:43 Lubos Lunak wrote:
>  In short: It's the Linux kernel's fault and it horribly sucks at this.
>  The  system becomes unresponsible because with memory full the system get
>  completely busy with just moving things between RAM and disk (swap), and
>  one of the things that is broken is that the kernel needs ages to realize
>  that something is wrong. As soon as it eventually manages to find out, O=
OM
>  (Out Of Memory) killer is started to pick a process to be killed, which
>  should remedy the situation. Here again it needs quite a while to make up
>  its mind and (especially with KDE) it often decides to kill completely
>  innocent vital processes first. It's been broken like this for ages and
>  apparently nobody capable of fixing it cares :(.
>=20
>  Before rebooting, you can try using SysRq+F to manually trigger the OOM=
=20
> killer, which should save some time. We have a hack in kdelibs to protect=
=20
> kdeinit (which often used to be the target), but basically there's no goo=
d=20
> workaround.
>=20
I'll try that next time - although getting it to accept any keyboard input =
is=20
extremely difficult.

>  If you know what process is likely to cause this, you can also run it wi=
th
>  a  memory limit set using 'ulimit'.
>=20
Can you give me a bit more detail?  If it *is* nspluginviewer that triggers=
=20
it, for instance, what do I need to do to set that limit?

The worst thing is that once the problem has been triggered you can't stop =
it. =20
I can use firefox or konqueror for a long time without any problem at all a=
s=20
long as nspluginviewer doesn't kick in.  I've started killing it as soon as=
 I=20
leave the page that needed it, but although the leak slows, it doesn't free=
 up=20
anying, and the leak seems to be simply slowed, not stopped.

Top, on this laptop, normally shows a healthyamount of RAM free with little=
 or=20
no swap usage.  At the moment, for instance, it shows

Mem:  	2061316k total, 	1989632k used, 	 71684k free
Swap:	1052216k total,	553948k used,	498168k free

It moved to this position in just the few minutes I was viewing an embedded=
=20
video.

At this I can go on using it for some time, but clearly the problem has bee=
n=20
triggered, so I'll need to keep an eye on it so that I can bail out before =
I=20
lose work.

Anne
=2D-=20
KDE Community Working Group
New to KDE4? - get help from http://userbase.kde.org

--nextPart15382529.vCPLkiFksl
Content-Type: application/pgp-signature; name=signature.asc 
Content-Description: This is a digitally signed message part.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEABECAAYFAksJETUACgkQbMErw/n0TZokfACePd1z3MTDN3epLNoHPf+oGPB2
2w0An3U2k8Jmvyvxo3r+//xQ7xku2lMK
=R08O
-----END PGP SIGNATURE-----

--nextPart15382529.vCPLkiFksl--

--===============0227967578==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

 
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<

--===============0227967578==--