This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


[Xen-devel] SMP / Multicore mode transition point

To: <xen-devel@xxxxxxxxxxxxxxxxxxx>
Subject: [Xen-devel] SMP / Multicore mode transition point
From: "Peter Teoh" <tthtlc@xxxxxxxxxxxxxx>
Date: Fri, 17 Aug 2007 13:04:57 +0800
Delivery-date: Thu, 16 Aug 2007 22:05:10 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxx
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
Several newbie question:
At which point in the source code (two function point - one for Xen, and another for linux-xen) is the processor properly setup to run in multiprocessor mode?  
VMX/SVM mode (specified through EFER_SVME bit, eg, in stop_svm() function) is always specific per-CPU, right?   Therefore, whenever the Xen make a transition from guest to host OS, must the entire set of CPUs all be switched at the same time?   Ie, can one CPU remain executing in Xen kernel, while another is executing in linux-xen kernel mode?
Answering the above, will help me to understand the next question better:
Logical speaking, the treatment for transition from guest to host should be asymmetrical to host to guest in pending IRQs.   Ie, in guest->host transition, even if there are pending IRQ in any of the CPUs, transition can still occurr, for security reasons.   But in host->guest transition, if there are pending IRQs, and if transition is allowed, it will lead to complicated re-entrancy problems.   So this has to be disallowed, right?
Xen-devel mailing list
<Prev in Thread] Current Thread [Next in Thread>
  • [Xen-devel] SMP / Multicore mode transition point, Peter Teoh <=