From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jens Axboe Subject: Re: regarding bug #5914 - fs corruption on SATA Date: Tue, 31 Jan 2006 09:47:36 +0100 Message-ID: <20060131084736.GP4215@suse.de> References: <20060127091002.GX4311@suse.de> <20060127092046.GY4311@suse.de> <58cb370e0601270146n40a5b249v53a736a8021c82f4@mail.gmail.com> <20060127095039.GA4311@suse.de> <1138390636.3017.2.camel@rousalka.dyndns.org> <1138406042.2953.0.camel@rousalka.dyndns.org> <20060130150843.GS4209@suse.de> <1138664026.3065.0.camel@rousalka.dyndns.org> <20060131072618.GH4215@suse.de> <5558.192.54.193.25.1138696781.squirrel@rousalka.dyndns.org> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from ns.virtualhost.dk ([195.184.98.160]:25349 "EHLO virtualhost.dk") by vger.kernel.org with ESMTP id S1750733AbWAaIqD convert rfc822-to-8bit (ORCPT ); Tue, 31 Jan 2006 03:46:03 -0500 Content-Disposition: inline In-Reply-To: <5558.192.54.193.25.1138696781.squirrel@rousalka.dyndns.org> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Nicolas Mailhot Cc: Bartlomiej Zolnierkiewicz , Tejun Heo , Jeff Garzik , Linux-ide On Tue, Jan 31 2006, Nicolas Mailhot wrote: >=20 > Le Mar 31 janvier 2006 08:26, Jens Axboe a =E9crit : > > On Tue, Jan 31 2006, Nicolas Mailhot wrote: > >> Le lundi 30 janvier 2006 =E0 16:08 +0100, Jens Axboe a =E9crit : > >> > On Sat, Jan 28 2006, Nicolas Mailhot wrote: > >> > > I can confirm today's patch is not OK. The same baseline with > >> > > yesterday's patch boot fine. > >> > > >> > Is this any better? > >> > >> This one seems to work fine. > > > > And you don't get "w/ FUA" messages from the problematic drives - a= nd > > your data appears safe? Just checking, we cannot take these corrupt= ion > > things lightly. >=20 > I didn't spend a lot of time on this, the build finished rather late > in the evening/night. What I can say is the dramatic breakage I had > before is gone and I don't think there was any error in dmesg (will > post it this evening if you want). With dm+raid when FUA broke things > it was difficult to miss (screenfulls of ATA/raid errors, fs > corruption on reboot, etc) >=20 > Now if you ask me if I did some heavy I/O to stress the system no I > didn't yet. If problems still lurk they are a lot less extensive than > they were before. I did trigger a full FS autorelabel so at least the > read part was tested a bit. Sounds like it works, if you saw the errors so quickly. Just trying to be absolutely sure, if you could check for the "w/ FUA" prints not bein= g there now it would confirm that the blacklist does its job. --=20 Jens Axboe