From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [kvm-devel] WARNING: at /usr/src/modules/kvm/mmu.c:390 account_shadowed() Date: Mon, 19 May 2008 17:30:52 +0300 Message-ID: <48318F1C.3030104@qumranet.com> References: <480ED76E.9040400@qumranet.com> <20080424171421.GA14073@dmt> <4812E30E.2030808@qumranet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Marcelo Tosatti , kvm@vger.kernel.org To: Thomas Cataldo Return-path: Received: from bzq-179-150-194.static.bezeqint.net ([212.179.150.194]:24809 "EHLO il.qumranet.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756974AbYESOay (ORCPT ); Mon, 19 May 2008 10:30:54 -0400 In-Reply-To: Sender: kvm-owner@vger.kernel.org List-ID: Thomas Cataldo wrote: > On Sat, Apr 26, 2008 at 10:08 AM, Avi Kivity wrote: > >> That strikes me as unlikely (though a valid scenario). >> >> An alternative explanation is that we're seeing a nonpae guest, so each page >> can be shadowed in two different roles (two quadrants for a pte page) or >> even four (for a pgd page). >> >> Thomas, are you running a 32-bit nonpae guest? >> >> > > We reproduiced this on a production machine, running with : > > [...] > kvm is version 66. > > The offending guest was a 32bit winXp (don't know if this uses pae or > not) with 384mo of ram. we use tap networking and vnc access. The vms > on this server are qcow2 files stored lvm (this might be relevant as > lots of vfs calls are in the traces) : > /dev/mapper/volGroupe-VMs on /mnt/machines_virtuelles type ext3 (rw) > /dev/mapper/volGroupe-VMs_snapshots on /mnt/VMs_snapshots type ext3 (rw) > > The command line used to start the vms looks like this : > /usr/bin/kvm -m 256 -hda /home/tom/kvm/root/root_vm.qcow2 -no-acpi > -net nic,model=e1000,macaddr=52:54:10:00:00:02 -net > tap,script=/etc/qemu-kvm.ifup,downscript=/etc/qemu-kvm.ifdown > -daemonize -monitor telnet:localhost:9002,server,nowait -vnc :2 -k fr > > I attached our dmesg output with all the warnings we grabbed. > From the trace it looks like a nonpae guest (paging32_...). Anyway this was fixed in kvm-68. -- error compiling committee.c: too many arguments to function