From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Randy.Dunlap" Subject: Re: Problems with SMP Date: Thu, 28 Oct 2004 08:13:18 -0700 Sender: linux-smp-owner@vger.kernel.org Message-ID: <41810C8E.3010708@osdl.org> References: <4180FC04.4010800@lenherr.name> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4180FC04.4010800@lenherr.name> List-Id: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: Thomas Lenherr Cc: linux-smp@vger.kernel.org Thomas Lenherr wrote: > Hi, > I'm using an Asus PC-DL and everything works fine, except I'm > experiencing troubles when copying files via a network fs (nfs and smb). > Because of that I wrote to the linux-fsdevel-list and to the lkml. > A exact description of the symptons may be read here: > http://bugme.osdl.org/show_bug.cgi?id=3608 > > My system had 2 P4 with HT activated, but a few days ago I deactivated > HT and instantly I experienced much less troubles (~2/9 files corrupted > via nfs instead of ~7/9). After this I compiled a kernel without > smp-support and the problem was gone. Got no more any corrupted files! > It's a bit weird that I got only troubles with network filesystems, but > that's the way it is ... > Please read the description given in the bug-report link given above for > further informations, but, as I said, when I posted this bug I hadn't > realized yet it got something to do with smp... > > I'd be very pleased if somebody could help me so that I could reactivate > my second processor :) You probably don't want to go thru the corruption again, but can you tell us if this happens with CONFIG_PREEMPT=y or n? There have been a few problems recently with PREEMPT not being used safely. If it happens with PREEMPT=y, can you test it with PREEMPT=n please. -- ~Randy