From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:52362)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1b0f6f-00086e-J3
	for qemu-devel@nongnu.org; Wed, 11 May 2016 21:08:02 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <famz@redhat.com>) id 1b0f6d-0001Ul-Ds
	for qemu-devel@nongnu.org; Wed, 11 May 2016 21:08:00 -0400
Date: Thu, 12 May 2016 09:07:51 +0800
From: Fam Zheng <famz@redhat.com>
Message-ID: <20160512010751.GA17594@ad.usersys.redhat.com>
References: <1462848659-28659-1-git-send-email-famz@redhat.com>
	<20160511114841.GE1683@redhat.com>
	<20160511115605.GD4524@noname.str.redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20160511115605.GD4524@noname.str.redhat.com>
Subject: Re: [Qemu-devel] [PATCH v4 00/27] block: Lock images when opening
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: Kevin Wolf <kwolf@redhat.com>
Cc: "Richard W.M. Jones" <rjones@redhat.com>, qemu-devel@nongnu.org, Max Reitz <mreitz@redhat.com>, Jeff Cody <jcody@redhat.com>, Markus Armbruster <armbru@redhat.com>, Eric Blake <eblake@redhat.com>, qemu-block@nongnu.org, stefanha@redhat.com, pbonzini@redhat.com, John Snow <jsnow@redhat.com>, berrange@redhat.com, den@openvz.org

On Wed, 05/11 13:56, Kevin Wolf wrote:
> Am 11.05.2016 um 13:48 hat Richard W.M. Jones geschrieben:
> > While I remember there is another concern that doesn't seem to be
> > addressed in this patch series.  What happens when a guest is paused?
> > I think exclusive locks should be converted to shared locks in that
> > case, since (only while the guest is paused) it _is_ safe to fish
> > around inside the guest's state.  Of course the lock must be restored
> > before the guest resumes.
> 
> I think it's still one of the cases where it's appropriate to require an
> "I know what I'm doing" flag. In paused guests, you still don't
> necessarily see the same contents as the guest does (because of guest
> caches). Apart from that, things like block jobs and NBD servers keep
> running even with a stopped VM.

Agreed. The external accessor can explicitly specify "lock-image=off" on its
command line to fish around.

Fam

> 
> The lock can only be dropped in cases where we can justify switching to
> BDRV_O_INACTIVE, and I don't think a simple stop/cont should do this.
> 
> Kevin