From: Juergen Gross Subject: Re: [Patch 1 of 2]: PV-domain SMP performance XEN-part Date: Mon, 22 Dec 2008 07:10:08 +0100 Message-ID: <494F2F40.8060401@fujitsu-siemens.com> References: <4948EFC5.8050701@fujitsu-siemens.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: George Dunlap Cc: "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org George Dunlap wrote: > The idea might be interesting, but this implementation needs some > work. Xen shouldn't be reading a guest-writable bit of memory to > determine if it's already delayed the guest; it needs to store that > information somewhere else. As it is, a rogue guest could extend its > time indefinitely by just writing 0 there every couple hundred > instructions. Thanks for finding this one! The correction is quite simple, I can just use sd->delay_desched for this purpose instead. > > It should probably be something taken care of in the individual > schedulers, rather than in the abstract scheduling code. > > At any rate, let's see if there's a significant performance difference > between this and yield-after-spinning-awhile, and then we can work on > the architecture. Agreed. Juergen -- Juergen Gross Principal Developer IP SW OS6 Telephone: +49 (0) 89 636 47950 Fujitsu Siemens Computers e-mail: juergen.gross@fujitsu-siemens.com Otto-Hahn-Ring 6 Internet: www.fujitsu-siemens.com D-81739 Muenchen Company details: www.fujitsu-siemens.com/imprint.html