Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S269277AbUJWAGB (ORCPT ); Fri, 22 Oct 2004 20:06:01 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S269252AbUJWAEf (ORCPT ); Fri, 22 Oct 2004 20:04:35 -0400 Received: from mail06.syd.optusnet.com.au ([211.29.132.187]:58777 "EHLO mail06.syd.optusnet.com.au") by vger.kernel.org with ESMTP id S269277AbUJWAD5 (ORCPT ); Fri, 22 Oct 2004 20:03:57 -0400 Message-ID: <41799FE0.1020403@kolivas.org> Date: Sat, 23 Oct 2004 10:03:44 +1000 From: Con Kolivas User-Agent: Mozilla Thunderbird 0.8 (X11/20040913) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Alastair Stevens Cc: linux-kernel@vger.kernel.org Subject: Re: 2.6.9-ck1: swap mayhem under UT2004 References: <200410222346.32823.alastair@altruxsolutions.co.uk> In-Reply-To: <200410222346.32823.alastair@altruxsolutions.co.uk> X-Enigmail-Version: 0.86.1.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig41CDF542A8BAEE16608479DE" Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2523 Lines: 58 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig41CDF542A8BAEE16608479DE Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Alastair Stevens wrote: > Con and others: I've been running 2.6.9-ck1 for a couple of days, and seem > to have hit on a major swapping issue.... > > My machine is a UP Athlon 2500+ with 512MB, and everything hums along > nicely under normal desktop usage. But when launching UT2004, it just > crawls and jerks like hell. At one point, it appeared to have frozen > completely, but I managed to switch to a text console to see what was > happening, and basically I'd hit a swap frenzy: kswapd was sucking 50% of > the CPU, fighting with the UT2004 process. > > My RAM appeared to be almost "full", with no cache/buffers, but only a few > hundred K of swap was actually being used, and this wasn't changing. > The kswapd frenzy carried on for at least a couple of minutes; then > suddenly everything went smooth again and the game played perfectly from > then on. > > This is definitely new behaviour; I've run every recent 2.6 kernel, with > and without the staircase scheduler patch (but not the full -ck), and > never had any problems before. Yes, I'm running the dratted Nvidia > driver, but that's not the issue as it's been loaded with every other > kernel. Switching back to 2.6.9-rc3 makes everything behave perfectly > again.... > > Any ideas? Any more info required? I've seen reports of this happening since 2.6.9 _even on mainline_. Something seems very sick with kswapd where it consumes massive amounts of cpu. Can you reproduce without any -ck patches? Others have already done so, but it seems to happen earlier with -ck. Con --------------enig41CDF542A8BAEE16608479DE Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.6 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org iD8DBQFBeZ/gZUg7+tp6mRURAhFCAJ9kQpWiMW9nJhZCltecYg5pOekwAACeL7Z0 T13ZVijDkBUkueOUd5+yVik= =FPtR -----END PGP SIGNATURE----- --------------enig41CDF542A8BAEE16608479DE-- - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/