From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Chen, Tiejun" Subject: Re: [v3][PATCH 1/1] xen:vtd: missing RMRR mapping while share EPT Date: Fri, 25 Jul 2014 16:43:07 +0800 Message-ID: <53D2189B.6080008@intel.com> References: <1406199618-21574-1-git-send-email-tiejun.chen@intel.com> <53D22E550200007800025DD8@mail.emea.novell.com> <53D2331D0200007800025E7B@mail.emea.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <53D2331D0200007800025E7B@mail.emea.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 , Yang Z Zhang Cc: Kevin Tian , "xen-devel@lists.xen.org" List-Id: xen-devel@lists.xenproject.org On 2014/7/25 16:36, Jan Beulich wrote: >>>> On 25.07.14 at 10:24, wrote: >> Jan Beulich wrote on 2014-07-25: >>>>>> On 24.07.14 at 19:12, wrote: >>>>>> From: Chen, Tiejun >>>>> Sent: Thursday, July 24, 2014 4:00 AM >>>>> >>>>> intel_iommu_map_page() does nothing if VT-d shares EPT page table. >>>>> So rmrr_identity_mapping() never create RMRR mapping but in some >>>>> cases like some GFX drivers it still need to access RMRR. >>>> >>>> this is not accurate. as long as RMRR is reported, it's always >>>> necessary to have the identity mapping there, not just needed in >>>> SOME CASES. :-) >>> >>> Actually (as also said before) I think "in some cases" is quite correct: >>> There's no guarantee that a device will actually access the region(s) >>> an RMRR may specify for it. A particular example would be the USB case >>> where iiuc these regions are needed only until legacy mode emulation gets >> turned off. >> >> Yes, even for Intel GFX, the RMRR is never really accessed. So we never saw >> any issue even without this patch. > > So then what's the patch good for? If it's just addressing a theoretical No, this is not a theoretical problem. I already reply to Yang, on BDW windows GFX driver really accesses this range now. Without this patch, the blue screen appears. Tiejun > issue, then this supports my intention to commit it only when the other > (theoretical only) problem also gets taken care of. > > Jan > > >