From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christophe Saout Subject: Re: kernel BUG at arch/x86/xen/mmu.c:1860! Date: Fri, 14 Jan 2011 16:33:52 +0100 Message-ID: <1295019232.30003.21.camel@leto.intern.saout.de> References: <20101227155314.GG3728@dumpdata.com> <20101228104256.GJ2754@reaktio.net> <1294153817.24719.3.camel@leto.intern.saout.de> <1294154342.24719.6.camel@leto.intern.saout.de> <1294166410.24719.11.camel@leto.intern.saout.de> <20110114152217.GB7455@dumpdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20110114152217.GB7455@dumpdata.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Konrad Rzeszutek Wilk Cc: xen-devel@lists.xensource.com, Teck Choon Giam List-Id: xen-devel@lists.xenproject.org Hi Konrad, > > (without strace, the bug is very common, about every third "lvcreate" > > command. Every lvcreate command triggers about 20 multipath > > Something must be busted with your udev rules for this to go in effect. > You should have multipathd process those and not run multipath for > every device mapper call. This is a standard Debian squeeze. I think udev is just calling those to figure out if a device should be handled by multipath or not (part of the device identification). I'd rather not fiddle with it, as it "works" even if Debian is shipping suboptimal rules. Anyhow, changing this would simply reduce the risk of running into the kernel issue, not remove it entirely, so I prefer the actual fix. :) Christophe