From mboxrd@z Thu Jan 1 00:00:00 1970 From: Gianluca Guida Subject: Re: INVVPID question. Date: Tue, 28 Apr 2009 13:49:47 +0100 Message-ID: <49F6FB6B.5020504@eu.citrix.com> References: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Keir Fraser Cc: "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org Keir Fraser wrote: > The instruction reference manual explicitly states that it is not guaranteed > to flush VPID 0 (but also says that it 'might'). Can't be relied on. Yes, this is why I was asking about current implementation. I've been recently playing with shadows and separate (guest/host) tlb flushing, and from what I've understood this is what is going on. Since we use it in vmx_flush_guest_tlbs(), we could perhaps find a way, in case tlb flushing optimizations are important to fix this (e.g. using vpid_sync_vcpu_all() when current is defined and we're sure there are no other vpid-tagged entries in the processor tlb). Gianluca > > -- Keir > > On 28/04/2009 13:18, "Gianluca Guida" wrote: > >> Hello, >> >> I have a simple question for the Intel EPT/VPID experts: do current >> implementation of invvpid flushes also VPID 0 when 2 is the first >> argument (as in vpid_sync_all() in Xen)? >> >> Thanks in advance, >> Gianluca >> >> _______________________________________________ >> Xen-devel mailing list >> Xen-devel@lists.xensource.com >> http://lists.xensource.com/xen-devel > >