From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756742AbYHYNf7 (ORCPT ); Mon, 25 Aug 2008 09:35:59 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754524AbYHYNfu (ORCPT ); Mon, 25 Aug 2008 09:35:50 -0400 Received: from il.qumranet.com ([212.179.150.194]:11822 "EHLO il.qumranet.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754511AbYHYNft (ORCPT ); Mon, 25 Aug 2008 09:35:49 -0400 Message-ID: <48B2B533.9070509@qumranet.com> Date: Mon, 25 Aug 2008 16:35:47 +0300 From: Avi Kivity User-Agent: Thunderbird 2.0.0.16 (X11/20080723) MIME-Version: 1.0 To: "Eric W. Biederman" CC: Ingo Molnar , linux-kernel@vger.kernel.org, kvm@vger.kernel.org Subject: Re: [PATCH] Fix emergency_restart (sysrq-b) with kvm loaded on Intel hosts References: <1219655506-27418-1-git-send-email-avi@qumranet.com> <20080825091508.GC9114@elte.hu> In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Eric W. Biederman wrote: > Ingo Molnar writes: > > >> * Avi Kivity wrote: >> >> >>> Enabling Intel VT has the curious side effect whereby the INIT signal >>> is blocked. Rather than comment on the wisdom of this side effect, >>> this patch adds an emergency restart reboot notifier, and modifies the >>> kvm reboot notifier to disable VT on emergency reboot. >>> > > Please no notifiers in emergency_restart. > > First emergency_restart is not supposed to work reliably it is a best effort tickle > the hardware thing. > > Second and more importantly whenever someone adds a notifier instead of a proper hook > to one a code path like this it seems like avoiding building a proper interface so > and I believe keeps us from getting all of the logic and the heuristics right. > > Why not just add a disable intel_vt if it is enabled call? > > We need to do it across all cpus. However, a reliable (and simpler) fix has emerged: reset via ACPI. That causes a true reset which VT does not block. -- error compiling committee.c: too many arguments to function