From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christian Tramnitz Subject: Re: xen-unstable pci passthrough Date: Fri, 04 Sep 2009 12:27:42 +0200 Message-ID: References: <715D42877B251141A38726ABF5CABF2C054B5B42EF@pdsmsx503.ccr.corp.intel.com> <715D42877B251141A38726ABF5CABF2C054B5B43E8@pdsmsx503.ccr.corp.intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <715D42877B251141A38726ABF5CABF2C054B5B43E8@pdsmsx503.ccr.corp.intel.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org Han, Weidong wrote: > Actually the 1:1 mapping patch is too tricky, we don't think it's suitable to be in upstream. The better way is the gfx vendors to remove this trick (maybe change their drivers) to make them virtualization friendly. I don't think that this is very likely. The userbase is to small (right now), not to talk about existing bugs that affect a larger user group that won't be fixed. Also Nvidia wants to see some extra $ for their premium products (capable of Multi OS), I don't think they will just change the drivers to make that feature available for every card. If anyhow possible we should try to get that "intelligence" into Xen/xen-qemu and not rely on the vendors. If tricky/untested/unstable it should be marked as such and its usage made optional, but I see a lot of potential in that direction. Best regards, Christian