From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Kiszka Subject: Re: Jan Kiszka to maintain kvm-kmod Date: Tue, 27 Oct 2009 12:20:03 +0100 Message-ID: <4AE6D763.7050108@siemens.com> References: <4AE565A9.4000907@redhat.com> <4AE5E19A.6010202@siemens.com> <4AE6CEDD.7070603@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: KVM list , Marcelo Tosatti To: Avi Kivity Return-path: Received: from thoth.sbs.de ([192.35.17.2]:19134 "EHLO thoth.sbs.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753212AbZJ0LUQ (ORCPT ); Tue, 27 Oct 2009 07:20:16 -0400 In-Reply-To: <4AE6CEDD.7070603@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: Avi Kivity wrote: > On 10/26/2009 07:51 PM, Jan Kiszka wrote: >> That forecast already promises the next rain > > That would be me. I think it's easily fixable, either by emulating user > return notifiers using preempt notifiers (sched_out -> on_user_return), > or by noping out user return notifiers completely and hooking the kvm > sched_out callback to call kvm_on_user_return. Of course, we won't see > the performance gain, but that's expected. I was playing with such an approach already. But something goes wrong (GPF in kvm_set_shared_msr) when booting Win7. Will check without kvm-kmod next. Jan -- Siemens AG, Corporate Technology, CT SE 2 Corporate Competence Center Embedded Linux