From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:45500)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <peterx@redhat.com>) id 1bk4NN-00080M-JI
	for qemu-devel@nongnu.org; Wed, 14 Sep 2016 03:12:58 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <peterx@redhat.com>) id 1bk4NH-00079M-Cu
	for qemu-devel@nongnu.org; Wed, 14 Sep 2016 03:12:56 -0400
Received: from mx1.redhat.com ([209.132.183.28]:52712)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <peterx@redhat.com>) id 1bk4NH-000796-7P
	for qemu-devel@nongnu.org; Wed, 14 Sep 2016 03:12:51 -0400
Date: Wed, 14 Sep 2016 15:12:43 +0800
From: Peter Xu <peterx@redhat.com>
Message-ID: <20160914071243.GM3776@pxdev.xzpeter.org>
References: <1473389864-19694-1-git-send-email-peterx@redhat.com>
	<1473389864-19694-3-git-send-email-peterx@redhat.com>
	<20160914055528.GM15077@voom.fritz.box>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20160914055528.GM15077@voom.fritz.box>
Subject: Re: [Qemu-devel] [PATCH v4 2/3] memory: introduce
 IOMMUOps.notify_flag_changed
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel/>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: David Gibson <david@gibson.dropbear.id.au>
Cc: qemu-devel@nongnu.org, mst@redhat.com, jasowang@redhat.com, vkaplans@redhat.com, alex.williamson@redhat.com, wexu@redhat.com, pbonzini@redhat.com, cornelia.huck@de.ibm.com, dgibson@redhat.com

On Wed, Sep 14, 2016 at 03:55:28PM +1000, David Gibson wrote:

[...]

> > -static void vtd_iommu_notify_started(MemoryRegion *iommu)
> > +static void vtd_iommu_notify_flag_changed(MemoryRegion *iommu,
> > +                                          IOMMUNotifierFlag old,
> > +                                          IOMMUNotifierFlag new)
> >  {
> >      VTDAddressSpace *vtd_as = container_of(iommu, VTDAddressSpace, iommu);
> 
> Shouldn't this have a sanity check that the new flags doesn't include
> MAP actions?

See your r-b for patch 3, thanks! So skipping this one.

[...]

> > +static void spapr_tce_notify_flag_changed(MemoryRegion *iommu,
> > +                                          IOMMUNotifierFlag old,
> > +                                          IOMMUNotifierFlag new)
> > +{
> > +    if (old == IOMMU_NOTIFIER_NONE && new == IOMMU_NOTIFIER_ALL) {
> > +        spapr_tce_notify_started(iommu);
> > +    } else if (old == IOMMU_NOTIFIER_ALL && new == IOMMU_NOTIFIER_NONE) {
> > +        spapr_tce_notify_stopped(iommu);
> > +    }
> 
> This is wrong.  We need to do the notify_start and stop actions if
> *any* bits are set in the new/old flags, not just if all of them are
> set.

Power should need both, right? I can switch all

  "== IOMMU_NOTIFIER_ALL"

into:

  "!= IOMMU_NOTIFIER_NONE"

in the next version if you like, but AFAICT they are totally the same.

> 
> I'd also prefer to see notify_started and notify_stopped folded into
> this function.

I can do that for v5.

Thanks,

-- peterx