From mboxrd@z Thu Jan 1 00:00:00 1970 From: Boris Ostrovsky Subject: Re: [PATCH] libxc: Defer initialization of start_page for HVM guests Date: Tue, 22 Dec 2015 09:48:59 -0500 Message-ID: <567962DB.3000704@oracle.com> References: <1450741524-27859-1-git-send-email-boris.ostrovsky@oracle.com> <56791AEB.2080409@citrix.com> <56795A33.4080407@oracle.com> <56795FE0.2080909@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset="windows-1252"; Format="flowed" Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: <56795FE0.2080909@citrix.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: =?UTF-8?Q?Roger_Pau_Monn=c3=a9?= , ian.jackson@eu.citrix.com, stefano.stabellini@eu.citrix.com, ian.campbell@citrix.com, wei.liu2@citrix.com Cc: jgross@suse.com, xen-devel@lists.xen.org List-Id: xen-devel@lists.xenproject.org On 12/22/2015 09:36 AM, Roger Pau Monn=E9 wrote: > El 22/12/15 a les 15.12, Boris Ostrovsky ha escrit: >> On 12/22/2015 04:42 AM, Roger Pau Monn=E9 wrote: >>> Also, why is everything done inside of alloc_magic_pages_hvm moved to >>> start_info_hvm? AFAICT we should only need to move the code that fills >>> the hvm_start_info struct, but the rest of the code already present in >>> alloc_magic_pages_hvm could be left as-is. >> That's what I was referring to in the commit message (and in the comment >> below it): most, if not all, of the stuff that alloc_magic_pages_hvm() >> currently does really has nothing to do with magic pages allocation. >> E.g. hvm_params settting, special pages initialization, etc. So I >> figured I could move all of this to a later point. But, as I said in the >> comment, I am not convinced that start_info() is the right place either. > Some of it has to do with magic pages IMHO (although I have to admit > this is probably a question of taste), for example it makes sense from > my PoV to allocate the hvm_info_page, the special pages (xenstore, > console...), the ioreq server pages and possibly the EPT identity map. > Moving the start_info stuff into it's own function also makes sense. Since there are many ways to separate this I'll wait for maintainers to = express their preference. (I don't know who is around now and I will disappear too on Friday until = January so this may have to wait until then). -boris