From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paolo Bonzini Subject: Re: [PATCH 2/2] xen:i386:pc_piix: create isa bridge specific to IGD passthrough Date: Mon, 03 Nov 2014 13:01:03 +0100 Message-ID: <54576E7F.3000901@redhat.com> References: <20140901060506.GA20186@redhat.com> <54042514.6090206@intel.com> <003AAFE53969E14CB1F09B6FD68C3CD478F16D2A@ORSMSX101.amr.corp.intel.com> <54277979.7070408@intel.com> <20140929100111.GA32459@redhat.com> <542A18BD.2060008@intel.com> <20141007072653.GB2424@redhat.com> <543622CC.6050807@intel.com> <20141012095021.GC9567@redhat.com> <544A0174.7000003@intel.com> <20141024134747.GA6024@redhat.com> <5451ED1E.1000300@intel.com> <5457335B.1000308@intel.com> <5457686E.7020601@redhat.com> <545768CC.3000903@intel.com> <54576B5A.50005@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <54576B5A.50005@intel.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: "Chen, Tiejun" , "Michael S. Tsirkin" Cc: "xen-devel@lists.xensource.com" , "qemu-devel@nongnu.org" List-Id: xen-devel@lists.xenproject.org On 03/11/2014 12:47, Chen, Tiejun wrote: > On 2014/11/3 19:36, Chen, Tiejun wrote: >> On 2014/11/3 19:35, Paolo Bonzini wrote: >>> On 03/11/2014 08:48, Chen, Tiejun wrote: >>>>>>>> I think the point was mostly to reserve 1f to prevent >>>>>>>> devices from using it. >>>>>>>> As we populate slots in order it doesn't seem to important ... >>>>>>> >>>>>>> If we populate slot at !1f GFX driver can't find this ISA bridge. >>>>>> >>>>>> Right, but I mean if no special options are used, 1f will typically >>>>>> stay free without any effort on our side. >>>>> >>>>> Yeah. >>>>> >>>>> Actually based on current info we know, seems 1f is just specific to >>>>> our >>>>> scenario :) So I always think we can occupy that. But Paolo and you >>>>> can >>>>> really determine this point. >>>> >>>> What's your idea? >>> >>> I do not have any objection to always occupying 1f for Xen IGD >>> passthrough. > > After I go back to look at this again, I hope you don't misunderstand > what Michael mean now. He was saying we don't need to create a new > separate machine specific to IGD passthrough. But that idea is just from > you :) It's difficult for me to follow, because xen_igd_passthrough_pc_hvm_init does not exist in the current tree. The patches seem good to me; I was assuming that the new machine type would call xen_igd_passthrough_pc_hvm_init, but apparently I'm wrong? Paolo