From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: Re: USB disk corruption on Xen 4.1.0 & Linux 2.6.38.7 Date: Tue, 31 May 2011 10:54:09 -0400 Message-ID: <20110531145409.GN14641@dumpdata.com> References: <8150663.20.1306842333642.JavaMail.root@zimbra.overnetdata.com> <9158059.26.1306843669556.JavaMail.root@zimbra.overnetdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <9158059.26.1306843669556.JavaMail.root@zimbra.overnetdata.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: Anthony Wright Cc: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org On Tue, May 31, 2011 at 01:07:49PM +0100, Anthony Wright wrote: > I build Xen 4.1.0 and linux 2.6.38.7 from source to do a DIY system based on Linux From Scratch (LFS). Ok.. Had you tried my #devel/next-2.6.38 branch in the past? > > When I try to read from a USB memory stick (around 64M big), I get corruption on the latest version of Xen. So this is from dom0 kernel, right? > > With Xen 3.x and linux 2.6.18 I can read the memory stick no problems. > > If I use the linux 2.6.38.7 kernel without a Xen hypervisor I can read the memory stick without problems. > > If I use the same setup, but this time introduce the Xen 4.1.0 hypervisor I get random blocks of sectors on the USB disk returning all zeros. Normally these blocks change each time I reboot, but are fixed once they've been established. On one occasion the blocks changed continuously each time I looked at the disk (I have a big file on the USB disk and I run md5sum before accessing it). What is the filesystem on the USB disk? > > Just to remove other possibilities I've checked the system's RAM and it's fine (I've seen similar problems before with dodgy RAM). If I try to access a hard disk or CDROM, everything is fine. > > This could be a problem with the way I've built the kernel or Xen, but it seems to suggest that there's a bug in Xen. > > I'm happy to try to isolate the problem further if it can't be reproduced elsewhere. My first suggestion would be to use a known good binary copy of xen 4.1.0 and/or linux 2.6.38.7 if anybody can supply one. Please provide serial log or dmesg output. > > thanks, > > Anthony Wright > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xensource.com > http://lists.xensource.com/xen-devel