From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <455E058C.1010404@domain.hid> Date: Fri, 17 Nov 2006 19:55:08 +0100 From: Jan Kiszka MIME-Version: 1.0 Subject: Re: [Xenomai-help] VM killer References: <455E025B.7060708@domain.hid> In-Reply-To: <455E025B.7060708@domain.hid> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig9B7D6825E08ED731F5E16244" Sender: jan.kiszka@domain.hid List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gilles Chanteperdrix Cc: xenomai@xenomai.org This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig9B7D6825E08ED731F5E16244 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Gilles Chanteperdrix wrote: > Daniel Schnell wrote: >> Hi, >> =20 >> I know this is somewhat not related to Xenomai. But as I encountered >> this while actually testing the memory allocation with Xenomai, maybe >> there is some common sense here, as there are many people having to do= >> with embedded systems, where something like that matters. >> =20 >> Everytime the system comes to its limits and some process uses up all >> the memory the vm systems kills (more or less randomly) one or more >> processes. Even system processes. I would really like to get rid of th= is >> behaviour, but having disabled CONFIG_OOM_KILLER inside the linux kern= el >> doesn't help. >> =20 >> I use ELDK4.0 with 2.4.25 Denx Kernel. >> =20 >> Any hints ? >=20 > There is no way the system can run properly if there is not enough RAM.= > Add some RAM to your board, or decrease the memory used by applications= =2E Having a means to avoid that your critical process gets killed accidentally because some uncritical one decided to suck up all system resources is, IMO, a very legitimate use case. On 2.6, where there is no CONFIG_OOM_KILLER, you can play with /proc//oom_adj and make your process "unattractive" to the OOM killer (set to -17). On older kernels you also had to include any parent processes in this protection. I think this was fixed meanwhile. But I wonder why there should be no effect of CONFIG_OOM_KILLER on 2.4. Forgot to replay the updated kernel binaries? Jan --------------enig9B7D6825E08ED731F5E16244 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFFXgWMniDOoMHTA+kRAs48AJ47TWBQwp4096Q7FCpS9re5H5jU5QCdEAac kdnafH4rt8H2CHxeoCsB92A= =NCvm -----END PGP SIGNATURE----- --------------enig9B7D6825E08ED731F5E16244--