From mboxrd@z Thu Jan  1 00:00:00 1970
From: Scott Wood <scottwood@freescale.com>
Subject: Re: [PATCH 2/2 v2] KVM: PPC: booke: Add watchdog emulation
Date: Tue, 17 Jul 2012 13:00:32 -0500
Message-ID: <5005A840.8070200@freescale.com>
References: <1341830087-12728-1-git-send-email-Bharat.Bhushan@freescale.com> <50044CF1.8070806@suse.de> <5004B995.8060303@freescale.com> <25F0D173-5EFA-4516-B478-F5E220E502CE@suse.de> <6A3DF150A5B70D4F9B66A25E3F7C888D03DCB08D@039-SN2MPN1-023.039d.mgd.msft <50055FEC.4020602@suse.de> <6A3DF150A5B70D4F9B66A25E3F7C888D03DCB328@039-SN2MPN1-023.039d.mgd.msft.net> <50057039.6070108@suse.de> <6A3DF150A5B70D4F9B66A25E3F7C888D03DCB508@039-SN2MPN1-023.039d.mgd.msft.net> <5005781E.1020502@suse.de> <5005925A.7090502@freescale.com> <50059820.7050305@suse.de>
Mime-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
Cc: Bhushan Bharat-R65777 <R65777@freescale.com>,
	Wood Scott-B07421 <B07421@freescale.com>,
	"<kvm-ppc@vger.kernel.org>" <kvm-ppc@vger.kernel.org>,
	"<kvm@vger.kernel.org>" <kvm@vger.kernel.org>,
	"<bharatb.yadav@gmail.com>" <bharatb.yadav@gmail.com>,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	Kumar Gala <galak@kernel.crashing.org>,
	Avi Kivity <avi@redhat.com>
To: Alexander Graf <agraf@suse.de>
Return-path: <kvm-ppc-owner@vger.kernel.org>
In-Reply-To: <50059820.7050305@suse.de>
Sender: kvm-ppc-owner@vger.kernel.org
List-Id: kvm.vger.kernel.org

On 07/17/2012 11:51 AM, Alexander Graf wrote:
> On 07/17/2012 06:27 PM, Scott Wood wrote:
>> Determining whether we're on final expiration is based only on the
>> previous value of TSR[ENW,WIS] when the timer expires.  We look at TCR
>> to determine whether we need to actually exit to QEMU (as opposed to a
>> final expiration with no action, which still should result in the timer
>> being stopped), but I don't see any problematic races there as long as
>> we don't try to update TCR from that context.
> 
> Yeah, what I was trying to say is that I would like to keep the TSR
> state unmodified for the final expiry case. If user space decides to not
> act upon it, it should be able to call into VCPU_RUN and have the same
> behavior as if the watchdog never expired.

I agree, it would be racy for QEMU to have to clear TSR if it's not
doing a full reset.  And that fact is making me wonder if I should have
listened to you and kept TSR owned by the vcpu thread. :-)

-Scott