From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752580Ab1IELov (ORCPT ); Mon, 5 Sep 2011 07:44:51 -0400 Received: from smtp.eu.citrix.com ([62.200.22.115]:32360 "EHLO SMTP.EU.CITRIX.COM" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752215Ab1IELon (ORCPT ); Mon, 5 Sep 2011 07:44:43 -0400 X-IronPort-AV: E=Sophos;i="4.68,333,1312156800"; d="scan'208";a="7603009" Date: Mon, 5 Sep 2011 12:52:42 +0100 From: Stefano Stabellini X-X-Sender: sstabellini@kaball-desktop To: Jeremy Fitzhardinge CC: Peter Zijlstra , "H. Peter Anvin" , Linus Torvalds , Ingo Molnar , the arch/x86 maintainers , Linux Kernel Mailing List , Nick Piggin , Avi Kivity , Marcelo Tosatti , KVM , Andi Kleen , Xen Devel , Jeremy Fitzhardinge , Stefano Stabellini , Keir Fraser Subject: Re: [PATCH 08/13] xen/pvticketlock: disable interrupts while blocking In-Reply-To: <4E614FBD.2030509@goop.org> Message-ID: References: <38bb37e15f6e5056d5238adac945bc1837a996ec.1314922370.git.jeremy.fitzhardinge@citrix.com> <1314974826.1861.1.camel@twins> <4E612EA1.20007@goop.org> <1314996468.8255.0.camel@twins> <4E614FBD.2030509@goop.org> User-Agent: Alpine 2.00 (DEB 1167 2008-08-23) MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org CC'ing Keir. On Fri, 2 Sep 2011, Jeremy Fitzhardinge wrote: > On 09/02/2011 01:47 PM, Peter Zijlstra wrote: > > On Fri, 2011-09-02 at 12:29 -0700, Jeremy Fitzhardinge wrote: > >>> I know that its generally considered bad form, but there's at least one > >>> spinlock that's only taken from NMI context and thus hasn't got any > >>> deadlock potential. > >> Which one? > > arch/x86/kernel/traps.c:nmi_reason_lock > > > > It serializes NMI access to the NMI reason port across CPUs. > > Ah, OK. Well, that will never happen in a PV Xen guest. But PV > ticketlocks are equally applicable to an HVM Xen domain (and KVM guest), > so I guess there's at least some chance there could be a virtual > emulated NMI. Maybe? Does qemu do that kind of thing? Xen knows how to inject NMIs to HVM guests, even though I am not sure if it is actually done in practice at the moment. However digging into the implementation details, it looks like virtual NMIs are not injected if blocking-by-STI (or blocking-by-MOV-SS), so we should be fine, even though I don't know if you actually want to rely on this: /* * We can only inject an NMI if no blocking by MOV SS (also, depending on * implementation, if no blocking by STI). If pin-based 'virtual NMIs' * control is specified then the NMI-blocking interruptibility flag is * also checked. The 'virtual NMI pending' control (available only in * conjunction with 'virtual NMIs') causes a VM exit when all these checks * succeed. It will exit immediately after VM entry if the checks succeed * at that point. * * Because a processor may or may not check blocking-by-STI when injecting * a virtual NMI, it will be necessary to convert that to block-by-MOV-SS * before specifying the 'virtual NMI pending' control. Otherwise we could * enter an infinite loop where we check blocking-by-STI in software and * thus delay delivery of a virtual NMI, but the processor causes immediate * VM exit because it does not check blocking-by-STI. */