From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Jan Beulich" Subject: Re: [v6][PATCH 2/2] xen:vtd: missing RMRR mapping while share EPT Date: Wed, 24 Sep 2014 08:54:08 +0100 Message-ID: <542294C002000078000380B7@mail.emea.novell.com> References: <541FB087.4080008@intel.com> <541FB7C3.9080608@intel.com> <541FFFC50200007800036C28@mail.emea.novell.com> <541FE65A.8070803@intel.com> <542017E80200007800036D1E@mail.emea.novell.com> <5420D357.1060202@intel.com> <5421803E0200007800037914@mail.emea.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Content-Disposition: inline List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Kevin Tian , Tiejun Chen Cc: Yang Z Zhang , "xen-devel@lists.xen.org" List-Id: xen-devel@lists.xenproject.org >>> On 24.09.14 at 02:28, wrote: > I'm a bit lost what's the exact regression here. This patch definitely is > not aimed > to solve all the problems, which is what Tiejun's another big series is > doing. it's > a step forward to allow GPU pass-through for Windows VM, w/o regression in > other areas (which are mostly broken today). However, seems your regression > is caused by patching other code: > > ---- > So I gave this a try on the one box I have which exposes RMRRs > (since those are for USB devices I also used your patch to drop > the USB special casing as done in your later patch series, and I > further had to fiddle with vtd_ept_page_compatible() in order to > get page table sharing to actually work on that box [I'll send the > resulting patch later]) > ---- > > If that's the case, it's not the regression typically defined. Sure - I had to do this extra patching to have a way to simulate the effect. I simply don't have another system where RMRRs would be associated with other than USB devices. The facts about the 2 patches here are: - they don't generally help (namely not for large enough guests) - they may yield devices unassignable that used to work fine Jan