From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stefan Subject: Re: Serious IDE problem still present in 2.6.14 Date: Thu, 10 Nov 2005 17:59:58 +0100 Message-ID: <43737C8E.90808@yahoo.de> References: <1130689871.6348.3.camel@montroll.chem.mcgill.ca> <4365511C.7040903@yahoo.de> <1130713716.6348.7.camel@montroll.chem.mcgill.ca> <4365596F.4060105@yahoo.de> <1130785344.5932.6.camel@montroll.chem.mcgill.ca> <58cb370e0511100534tead9e71u189666789b3f4da1@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from smtp013.mail.yahoo.com ([216.136.173.57]:32606 "HELO smtp013.mail.yahoo.com") by vger.kernel.org with SMTP id S1750899AbVKJRA7 (ORCPT ); Thu, 10 Nov 2005 12:00:59 -0500 In-Reply-To: <58cb370e0511100534tead9e71u189666789b3f4da1@mail.gmail.com> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Bartlomiej Zolnierkiewicz Cc: David.Ronis@mcgill.ca, Mark Hahn , linux-ide@vger.kernel.org Bartlomiej Zolnierkiewicz wrote: >I would suggest doing binary search on -rc and snapshot releases. > >http://kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.14-rc* >http://kernel.org/pub/linux/kernel/v2.6/snapshots/old/patch-2.6.14-* > >Otherwise we can only guess what is wrong. > >Moreover it doesn't look like an IDE problem the real performance >regression is in cached reads not the buffered ones. > > Exactly, that's why I advised him to send another mail to the linux-kernel list, since I think it is related to some of the acpi changes from 2.6.12 to >=2.6.13. If he switches off acpi then performance is back to normal and since there were basically no changes in the ide code or code for his controller it must be related to other changes. >On 10/31/05, David Ronis wrote: > > >>Hi Stefan, >> >>I tried removing the modules you suggested and reran hdparm -tT /dev/hda >>after each. No difference. I also played a bit more by going to >>single-user mode and killing off various things (like udevd). I turned >>off networking and removed the ethernet driver--still no change. What >>seemed to make a difference was to turn of the firewall and remove all >>the ipt_ drivers, although the performance was still about 4 times less >>than what I get with the 2.6.13.6 kernel. >> >> > > > ___________________________________________________________ Gesendet von Yahoo! Mail - Jetzt mit 1GB Speicher kostenlos - Hier anmelden: http://mail.yahoo.de