Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752583AbaJ1NDL (ORCPT ); Tue, 28 Oct 2014 09:03:11 -0400 Received: from cantor2.suse.de ([195.135.220.15]:46169 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751062AbaJ1NDK (ORCPT ); Tue, 28 Oct 2014 09:03:10 -0400 Message-ID: <544F940B.8090707@suse.com> Date: Tue, 28 Oct 2014 14:03:07 +0100 From: Juergen Gross User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.1.0 MIME-Version: 1.0 To: Andrew Cooper , David Vrabel , Ian Campbell CC: boris.ostrovsky@oracle.com, xen-devel@lists.xensource.com, linux-kernel@vger.kernel.org Subject: Re: [Xen-devel] [PATCH 0/2] xen: Switch to virtual mapped linear p2m list References: <1414421551-31555-1-git-send-email-jgross@suse.com> <544E61DD.8050305@citrix.com> <544F22E4.9040706@suse.com> <1414489898.23883.25.camel@citrix.com> <544F83C1.3010402@citrix.com> <544F8713.9040803@suse.com> <544F8E89.6070006@citrix.com> <544F8F3A.3030505@citrix.com> <544F8FBB.2010709@citrix.com> <544F902C.8040502@citrix.com> In-Reply-To: <544F902C.8040502@citrix.com> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 10/28/2014 01:46 PM, Andrew Cooper wrote: > On 28/10/14 12:44, David Vrabel wrote: >> On 28/10/14 12:42, Andrew Cooper wrote: >>> On 28/10/14 12:39, David Vrabel wrote: >>>> On 28/10/14 12:07, Juergen Gross wrote: >>>>> Okay, back to the original question: is the (up to) 64 MB virtual >>>>> mapping of the p2m list on 32-bit pv domains a problem or not? >>>> I think up-to 64 MiB of vmalloc area is fine. The vmalloc space can be >>>> increased with a command line option in the unlikely event that there >>>> are domUs that would be affected. >>>> >>>>> If yes, the virtual mapped linear p2m list could still be used on >>>>> 64 bit domains, paving the way for support of more than 512 GB of >>>>> domain memory. OTOH having to keep the p2m tree coding alive isn't >>>>> my favorite solution... >>>> Having to keep both the tree and linear p2m code would be awful. Let's >>>> not do this! >>>> >>>> David >>> How is the toolstack expected to find and mutate this p2m on migrate? >>> The toolstack does not use guest pagetables. >> The p2m code maintains two trees. One for use by the guest and one for >> use by the toolstack. This series replaces the first tree with the >> linear array (at least that's how I understand it -- I've not reviewed >> the series in detail yet). > > Ah - fine (if that is how it works). That's the plan, yes. Juergen -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/