From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261798AbUEFOXG (ORCPT ); Thu, 6 May 2004 10:23:06 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262438AbUEFOXF (ORCPT ); Thu, 6 May 2004 10:23:05 -0400 Received: from piggy.rz.tu-ilmenau.de ([141.24.4.8]:55008 "EHLO piggy.rz.tu-ilmenau.de") by vger.kernel.org with ESMTP id S261798AbUEFOW4 (ORCPT ); Thu, 6 May 2004 10:22:56 -0400 Date: Thu, 6 May 2004 16:22:47 +0200 From: "Mario 'BitKoenig' Holbe" To: Henrik Persson Cc: linux-kernel@vger.kernel.org Subject: Re: Strange DMA-errors and system hang with Promise 20268 Message-ID: <20040506142247.GC22719@darkside.22.kls.lan> Mail-Followup-To: Mario 'BitKoenig' Holbe , Henrik Persson , linux-kernel@vger.kernel.org References: <1078602426.16591.8.camel@vega> <200405052339.i45NdXsx003369@darkside.22.kls.lan> <1083849053.6994.10.camel@vega> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1083849053.6994.10.camel@vega> User-Agent: Mutt/1.5.6i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thu, May 06, 2004 at 03:10:54PM +0200, Henrik Persson wrote: > On Thu, 2004-05-06 at 01:39, Mario 'BitKoenig' Holbe wrote: > > just to verify things: what is your io_32bit set to? > io_32bit 0 0 3 rw Well, it was a chance :) However, I had io_32bit=1 but I also had a freeze even after setting it back to 0. > And well. I don't have any problems nowadays. Not a freeze since I sent > that mail. Strange indeed. I found a few things: 1st: it happens when the disk is loaded: best way to freeze the system here is a find over a big subtree (something around 60.000 files in lots of directories, ext2), better some of them in parallel and even better some find -print0 | xargs -0 cat > /dev/null. But even there: it doesnt happen all time, the chances just grow up. 2nd: it seems, the disk spins down short before the error appears. At least all time when I was in front of the machine when the problem appeared, I heard the disk spinning down - just as if it was gone to standby mode. But it was definitely loaded before (I hear the head seeks and then the spindown). And no, I dont have any automated spindown like noflushd or something like that. CC:ing to lkm again. Probably this helps somehow. regards, Mario -- [mod_nessus for iauth] "scanning your system...found depreciated OS...found hole...installing new OS...please reboot and reconnect now"