From mboxrd@z Thu Jan 1 00:00:00 1970 From: Keir Fraser Subject: Re: Re: Xen-unstable save error Date: Tue, 22 Jun 2010 11:01:26 +0100 Message-ID: References: <4C2078E202000078000078BB@vpn.id2.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <4C2078E202000078000078BB@vpn.id2.novell.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: Jan Beulich , Michal Novotny Cc: "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org On 22/06/2010 07:48, "Jan Beulich" wrote: >>>> On 22.06.10 at 08:23, Michal Novotny wrote: >> Jan, as we were solving this with Keir on the list I was unable to save >> the HVM guest properly and reverting your c/s 21615 helped and with your >> patch reverted it was working fine. Any idea why there were so type >> failures for all the pages of HVM guest? > > Not really, and I think you will need to find out by instrumenting the > hypercall implementation (XEN_DOMCTL_getpageframeinfo{2,3}) - > really, type failures should only happen for grant frames and I/O > memory pages (as you can see in the code, for the function to > report XTAB either the MFN must be invalid, must be a Xen frame, > or it must be impossible to obtain a reference to the page - all of > which shouldn't hold for the majority of a HVM guest's pages). Michal, is it possible you are running new tools against a slightly older hypervisor? The tools changes had a corresponding, non-version-checked, modification to a hypercall. Without running the modified hypervisor, HVM save would indeed behave very oddly! -- Keir