This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


Re: [Xen-devel] double or triple access to ext2, ext3 or other partition

To: Mark Williamson <maw48@xxxxxxxxxxxxxxxx>
Subject: Re: [Xen-devel] double or triple access to ext2, ext3 or other partitions
From: Luke Kenneth Casson Leighton <lkcl@xxxxxxxx>
Date: Tue, 23 Nov 2004 21:06:36 +0000
Cc: xen-devel@xxxxxxxxxxxx
Delivery-date: Tue, 23 Nov 2004 20:56:46 +0000
Envelope-to: xen+James.Bulpin@xxxxxxxxxxxx
In-reply-to: <Pine.LNX.4.60.0411231724580.3258@xxxxxxxxxxxxxxxxxxxxxx>
List-archive: <http://sourceforge.net/mailarchive/forum.php?forum=xen-devel>
List-help: <mailto:xen-devel-request@lists.sourceforge.net?subject=help>
List-id: List for Xen developers <xen-devel.lists.sourceforge.net>
List-post: <mailto:xen-devel@lists.sourceforge.net>
List-subscribe: <https://lists.sourceforge.net/lists/listinfo/xen-devel>, <mailto:xen-devel-request@lists.sourceforge.net?subject=subscribe>
List-unsubscribe: <https://lists.sourceforge.net/lists/listinfo/xen-devel>, <mailto:xen-devel-request@lists.sourceforge.net?subject=unsubscribe>
References: <20041123171312.GC6250@xxxxxxxx> <Pine.LNX.4.60.0411231724580.3258@xxxxxxxxxxxxxxxxxxxxxx>
Sender: xen-devel-admin@xxxxxxxxxxxxxxxxxxxxx
User-agent: Mutt/
On Tue, Nov 23, 2004 at 05:30:04PM +0000, Mark Williamson wrote:
> >2) at every point in the ext2 and ext3 filesystem drivers where
> >SMP-driven locking is performed, add in xen-semaphore usage and locking
> >(as well / instead).
> It's not quite that simple: the ext driver in each domain will be caching 
> data related to its filesystem changes in memory.  In an SMP system, both 
> CPUs can see this cached data in memory, whereas separate virtual machines 
> cannot see what each other are caching.

 oh drat.

> Mounting the same partition is already possible using a cluster filesystem 
> (like GFS).  Cluster filesystems handle the extra locking correctly 
> because they're designed for use with multiple hosts sharing a disk (e.g. 
> over a SAN).  They'll even work with multiple writers.
 oooooo :)

> It is also possible to export any filesystem via NFS and have it mounted 
> by multiple domains or to have copy-on-write semantics (e.g. using LVM) so 
> that domains never see each others changes (thus circumventing the 
> problem).

 in that case, the NFS server becomes the single point of failure should
 it become attacked / compromised.

 thank you for your advice about gfs, i will most definitely investigate



SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now. 
Xen-devel mailing list

<Prev in Thread] Current Thread [Next in Thread>