From: Ian Campbell Subject: Re: [PATCH 2/2] xen: make start_info x86 specific. Date: Fri, 15 Feb 2013 14:06:36 +0000 Message-ID: <1360937196.31407.76.camel@zakaz.uk.xensource.com> References: <1360851380.20449.397.camel@zakaz.uk.xensource.com> <1360851590-2539-2-git-send-email-ian.campbell@citrix.com> <511E3FAB02000078000BEAD2@nat28.tlf.novell.com> <1360934649.31407.44.camel@zakaz.uk.xensource.com> <511E48FC02000078000BEB2E@nat28.tlf.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <511E48FC02000078000BEB2E@nat28.tlf.novell.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: Jan Beulich Cc: "Tim (Xen.org)" , "xen-devel@lists.xen.org" , "Keir (Xen.org)" , Konrad Rzeszutek Wilk , Stefano Stabellini List-Id: xen-devel@lists.xenproject.org On Fri, 2013-02-15 at 13:41 +0000, Jan Beulich wrote: > >>> On 15.02.13 at 14:24, Ian Campbell wrote: > > All I actually care about is the removal of struct start_info from the > > canonical copy of the hypervisor ABI (that is xen/include/public) for > > ARM. > > But as already said - to me this looks wrong. The structure has > generally useful fields (and unless ARM uses _none_ of them, This is right, ARM uses none of them, at the hypervisor ABI level at least. There's actually no way to pass a start info to an ARM guest because they are launched via the native code paths. > then it must have some other way to communicate the respective > stuff, in which case I'd ask why a second mechanism got invented > when there already was one available), and even more so from > the abstract PV perspective. ARM uses HVM params (e.g. the xenstore and console ring + evtchn) and XENFEAT_dom0 etc (for privilegedness). I guess it is obvious why HVM params were needed on HVM and you invented XENFEAT_dom0 IIRC :-) Ian.