From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Cooper Subject: Re: Buggy interaction of live migration and p2m updates Date: Fri, 21 Nov 2014 10:24:50 +0000 Message-ID: <546F12F2.3020809@citrix.com> References: <546E32BB.8090909@citrix.com> <1416562990.26869.10.camel@citrix.com> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============1748438345909801221==" Return-path: In-Reply-To: <1416562990.26869.10.camel@citrix.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Ian Campbell Cc: Juergen Gross , Wei Liu , Ian Jackson , Tim Deegan , Xen-devel List , David Vrabel , Jan Beulich , Shriram Rajagopalan , Hongyang Yang List-Id: xen-devel@lists.xenproject.org --===============1748438345909801221== Content-Type: multipart/alternative; boundary="------------040309030801080206060405" --------------040309030801080206060405 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable On 21/11/14 09:43, Ian Campbell wrote: > On Thu, 2014-11-20 at 18:28 +0000, Andrew Cooper wrote: >> Realistically, this means no updates to the >> p2m at all, due to several potential race conditions. > From the rest of the mail it seems as if you are talking primarily abou= t > changes to the p2m *structure*, i.e. which guest frames contain the p2m= > pages, rather than changes to the p2m entries themselves. Is that > correct? I cover both, although the structure changes are the more obvious, and more complicated to fix. There are race conditions in both existing implementation between a p2m/m2p update and the toolstack sampling the dirty bitmap where stale frames don't get resent. > > I don't see any (explicit) mention of the pfn_to_mfn_frame_list_list > here, where does that fit in? > It is referenced several times, although not by its exact name. ~Andrew --------------040309030801080206060405 Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: 8bit
On 21/11/14 09:43, Ian Campbell wrote:
On Thu, 2014-11-20 at 18:28 +0000, Andrew Cooper wrote:

Realistically, this means no updates to the
p2m at all, due to several potential race conditions.

From the rest of the mail it seems as if you are talking primarily about
changes to the p2m *structure*, i.e. which guest frames contain the p2m
pages, rather than changes to the p2m entries themselves. Is that
correct?

I cover both, although the structure changes are the more obvious, and more complicated to fix.

There are race conditions in both existing implementation between a p2m/m2p update and the toolstack sampling the dirty bitmap where stale frames don't get resent.


I don't see any (explicit) mention of the pfn_to_mfn_frame_list_list
here, where does that fit in?

It is referenced several times, although not by its exact name.

~Andrew
--------------040309030801080206060405-- --===============1748438345909801221== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel --===============1748438345909801221==--