From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: Re: [PATCH 2 of 3] apic: remove 'enabled_via_apicbase' variable Date: Wed, 18 May 2011 16:45:37 -0400 Message-ID: <20110518204537.GB7647@dumpdata.com> References: <20110518185339.GD14013@dumpdata.com> <4DD42D9E.7030700@citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <4DD42D9E.7030700@citrix.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: Andrew Cooper Cc: "xen-devel@lists.xensource.com" List-Id: xen-devel@lists.xenproject.org > >So if you don't have x2apic, then it is wrong to disable the LAPIC mode? > >What about older hardware? > I guess I wasn't very clear in my description. In older hardware > without x2apic, it is correct to simply twiddle the ENABLE bit in > the APICBASE MSR. However, with x2apic mode enabled, setting the > ENABLE bit from 1 to 0 while leaving the EXTD bit set will result in > a protection fault which will propagate to a general protection > fault the same codepath will be called in the fault handler. As a > result, the current code in disable_local_APIC will result in a GPF > if the BIOS boots with LAPICs disabled (fine as per the spec for > compatibility) and xen decided to take advantage of x2apic mode. Ok. You might also point to the appropiate section of the x2APIC spec about this. I think it is 2.7.1.2