From mboxrd@z Thu Jan  1 00:00:00 1970
From: Marcelo Tosatti <mtosatti@redhat.com>
Subject: Re: [patch 4/5] KVM: MMU: reload request from GET_DIRTY_LOG path
Date: Tue, 1 Jul 2014 14:50:10 -0300
Message-ID: <20140701175010.GA10927@amt.cnet>
References: <20140618231203.846608908@amt.cnet>
 <20140618231521.718959400@amt.cnet>
 <20140619081719.GD10948@minantech.com>
 <20140619184031.GC32410@amt.cnet>
 <20140620104610.GD20764@minantech.com>
 <20140630205902.GB26566@amt.cnet>
 <20140701062719.GK18167@minantech.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: kvm@vger.kernel.org, ak@linux.intel.com, pbonzini@redhat.com,
	xiaoguangrong@linux.vnet.ibm.com, avi@cloudius-systems.com
To: Gleb Natapov <gleb@minantech.com>
Return-path: <kvm-owner@vger.kernel.org>
Received: from mx1.redhat.com ([209.132.183.28]:11505 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1756327AbaGARuf (ORCPT <rfc822;kvm@vger.kernel.org>);
	Tue, 1 Jul 2014 13:50:35 -0400
Content-Disposition: inline
In-Reply-To: <20140701062719.GK18167@minantech.com>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On Tue, Jul 01, 2014 at 09:27:19AM +0300, Gleb Natapov wrote:
> > The vcpu will never see a read-only spte because the VM-exit (due to
> > IPI) guarantees vcpu is outside of guest mode _before_ it is write
> > protected.
> Right. Now I see why you absolutely have to send IPI in mmu_reload_pinned_vcpus()
> before marking pte as read only. And kvm->mmu_lock is what will prevent vcpu from
> re-entering guest mode again before pte is marked read only, right?

Yes.

> > So i ask you: do you still hold the "current approach should be
> > improved" position ?
> > 
> As I said IMO what I proposed is much simpler and not as tricky as what you have here.
> It also has an advantage of not slowing down next guest entry after GET_DIRTY_LOG because
> it does not require mmu reload and page_faulting in pinned pages.

Ok sure.