From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Cooper Subject: Re: VT-d flush timeout Date: Thu, 25 Sep 2014 11:56:18 +0100 Message-ID: <5423F4D2.60605@citrix.com> References: <53F2042E02000078000BAAD6@mail.emea.novell.com> <53F358F702000078000BAC69@mail.emea.novell.com> <5423F4A30200007800038CB2@mail.emea.novell.com> <5423F229.3020304@citrix.com> <542410A90200007800038E3B@mail.emea.novell.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <542410A90200007800038E3B@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 , Aravind Gopalakrishnan , suravee.suthikulpanit@amd.com Cc: LiangX Z Li , Yang ZZhang , Kevin Tian , Eddie Dong , "xen-devel@lists.xen.org" List-Id: xen-devel@lists.xenproject.org On 25/09/14 11:55, Jan Beulich wrote: >>>> On 25.09.14 at 12:44, wrote: >> The AMD code will wait for a period of time for the COMMAND_WAIT to >> start, but not panic() on a timeout. >> >> By my reading of the code, amd_iommu_flush_iotlb() can complete without >> confirming that the iotlbs have actually been flushed, and all that is >> left behind is an AMD_IOMMU_DEBUG() message. > Now that you say this I recall having stumbled across this questionable > behavior in the past (and perhaps even more than once). Certainly > something for our AMD IOMMU maintainers to look into... > > Jan > Agreed, although for now I think it is safer to leave ATS disabled. ~Andrew