From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: Re: [Xen-devel] [PATCH 6/7] xen/setup: Make dom0_mem=XGB behavior be similar to classic Xen kernels. Date: Mon, 9 Apr 2012 17:33:29 -0400 Message-ID: <20120409213329.GC12783@phenom.dumpdata.com> References: <1333139850-28456-1-git-send-email-konrad.wilk@oracle.com> <1333139850-28456-7-git-send-email-konrad.wilk@oracle.com> <4F7ABBC4.4050106@citrix.com> <4F7AE321020000780007C456@nat28.tlf.novell.com> <20120406210117.GB26309@phenom.dumpdata.com> <4F831ED7020000780008228C@nat28.tlf.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <4F831ED7020000780008228C@nat28.tlf.novell.com> Sender: linux-kernel-owner@vger.kernel.org To: Jan Beulich Cc: xen-devel@lists.xensource.com, david.vrabel@citrix.com, linux-kernel@vger.kernel.org List-Id: xen-devel@lists.xenproject.org On Mon, Apr 09, 2012 at 05:39:35PM +0100, Jan Beulich wrote: > >>> Konrad Rzeszutek Wilk 04/06/12 11:06 PM >>> > >On Tue, Apr 03, 2012 at 10:46:41AM +0100, Jan Beulich wrote: > >> >>> On 03.04.12 at 10:58, David Vrabel wrote: > >> > With your new behaviour it will no longer possible to specify an > >> > unlimited balloon but a limited number of initial pages. This is > >> > behaviour that Jan said he used. > >> > >> An unlimited balloon was never possible afaict (as that would have > >> implied setting up an "infinite" number of struct page instances at > >> boot time. > >> > >> What I'm using is "dom0_mem=-M" together with the kernel > >> option "mem=G", such that max-balloon > initial alloc (usually > >> I set max-balloon to approximately the amount of memory in the > >> system, so the upper limit is "infinite" in the sense that I can't go > >> higher anyway, but it's not truly infinity). > > > >Couldn't you do the same thing with 'dom0_mem=X,max:Y' > > That would be possible (albeit not exactly identical in behavior). But my > main point in the discussion was to not modify existing behavior without > actual need to (including the desire to not have to modify dozens of > command lines). Right. I don't want to modify the hypervisor - my goal is to bring the pvops kernel in line with how XenOLinux does it.