From owner-freebsd-current@FreeBSD.ORG Mon Aug 30 13:52:32 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 040C616A4CE; Mon, 30 Aug 2004 13:52:32 +0000 (GMT) Received: from av5-1-sn4.m-sp.skanova.net (av5-1-sn4.m-sp.skanova.net [81.228.10.112]) by mx1.FreeBSD.org (Postfix) with ESMTP id B36BB43D4C; Mon, 30 Aug 2004 13:52:31 +0000 (GMT) (envelope-from daniel_k_eriksson@telia.com) Received: by av5-1-sn4.m-sp.skanova.net (Postfix, from userid 502) id 1CBA737E5A; Mon, 30 Aug 2004 15:52:31 +0200 (CEST) Received: from smtp2-2-sn4.m-sp.skanova.net (smtp2-2-sn4.m-sp.skanova.net [81.228.10.182]) by av5-1-sn4.m-sp.skanova.net (Postfix) with ESMTP id 0C98537E43; Mon, 30 Aug 2004 15:52:31 +0200 (CEST) Received: from gadget (h130n1fls11o822.telia.com [213.64.66.130]) by smtp2-2-sn4.m-sp.skanova.net (Postfix) with ESMTP id C18DD37E42; Mon, 30 Aug 2004 15:52:30 +0200 (CEST) From: "Daniel Eriksson" To: "'Lukas Ertl'" , "'Craig Boston'" , Date: Mon, 30 Aug 2004 15:52:31 +0200 Organization: Home Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Office Outlook, Build 11.0.6353 In-Reply-To: <20040827012953.T556@korben.in.tern> X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.2180 Thread-Index: AcSLxU/8rJMyEu00TqmQagezZGSQFQC0H+oQ Subject: RE: PLEASE TEST: IPI deadlock avoidance patch X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 30 Aug 2004 13:52:32 -0000 Lukas Ertl wrote: > I've identified and fixed (hopefully) a source of possible > data corruption on RAID5 plexes three days ago. You need > rev. 1.6 of geom_vinum_raid5.c. I have now tested with 6-CURRENT compiled from sources dated 2004.08.29.21.00.00 (which includes your patch), and I must unfortunately report that it does not fix the problem. I tested by duplicating a bunch (9GB+) of binary files with known crc checksums on the same filesystem, and I ended up with ~10% corrupt files. Running the exact same test under a kernel compiled without SMP worked without any problems. (Testbed: SMP machine with 4 SATA discs in a gvinum RAID-5 array) I have managed to secure some more testing time on the server today, so I will be trying this with regular vinum later to see if that makes a difference. I'll report back here on the list. /Daniel Eriksson