From mboxrd@z Thu Jan 1 00:00:00 1970 From: George Dunlap Subject: Re: [PATCH 08 of 10 [RFC]] xl: Introduce First Fit memory-wise placement of guests on nodes Date: Thu, 3 May 2012 11:16:15 +0100 Message-ID: <4FA25AEF.8010403@eu.citrix.com> References: <31163f014d8a2da9375f.1334150275@Solace> <4FA0051F.6020909@eu.citrix.com> <1335976251.2961.123.camel@Abyss> <1336007032.874.12.camel@Abyss> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1336007032.874.12.camel@Abyss> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Dario Faggioli Cc: Andre Przywara , Ian Campbell , Stefano Stabellini , Juergen Gross , Ian Jackson , "xen-devel@lists.xen.org" , Jan Beulich List-Id: xen-devel@lists.xenproject.org On 03/05/12 02:03, Dario Faggioli wrote: > Actually, thinking more about this, I'm not even sure I can do what I > stated above. In fact, I don't think I can assume the number of CPUs > each node is made up of to be known, without actually checking it for > the specific node(s) I want to try using. What if some CPUs are off-line > and stuff like that? Hmm, yeah, off-line cpus are certainly going to be an issue. If each node has an equal number of cores, we should at least be able to find a *minimum* number of nodes, and then increase it from there; then in the common case (where there are no offline cpus), we'll only have a single pass. But I think the priority right now is getting something useable for the 4.2 release; so I think getting to a reasonable functionality quickly (i.e., with a minimum amount of effort from you) makes sense. If that means leaving it as is for the moment, that's fine. This won't be a hot path, so a little bit of inefficiency should be acceptable. We can always come back and revisit it later. -George