From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:38997)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1Zlvvx-000051-Nl
	for qemu-devel@nongnu.org; Tue, 13 Oct 2015 05:31:55 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1Zlvvw-0006LT-Ry
	for qemu-devel@nongnu.org; Tue, 13 Oct 2015 05:31:49 -0400
Date: Tue, 13 Oct 2015 11:31:42 +0200
From: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20151013093142.GE4906@noname.str.redhat.com>
References: <1444650651-26227-1-git-send-email-famz@redhat.com>
	<1444650651-26227-6-git-send-email-famz@redhat.com>
	<561BC33A.4030108@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <561BC33A.4030108@redhat.com>
Subject: Re: [Qemu-devel] [PATCH v2 05/12] block: Introduce "drained
	begin/end" API
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: Paolo Bonzini <pbonzini@redhat.com>
Cc: Stefan Hajnoczi <stefanha@redhat.com>, Fam Zheng <famz@redhat.com>, qemu-devel@nongnu.org, qemu-block@nongnu.org

Am 12.10.2015 um 16:27 hat Paolo Bonzini geschrieben:
> 
> 
> On 12/10/2015 13:50, Fam Zheng wrote:
> > +void bdrv_drained_begin(BlockDriverState *bs)
> > +{
> > +    if (bs->quiesce_counter++) {
> > +        return;
> > +    }
> > +    aio_disable_external(bdrv_get_aio_context(bs));
> > +    bdrv_drain(bs);
> > +}
> 
> I think bdrv_drain should be called unconditionally, i.e. before the
> "if".  This should also solve Kevin's doubt about new allocating write
> request reenabling the timer: any write request from the drained section
> happens normally, until you get a nested drain request and then the
> callback completes the requests.

This would mean that once you've sent an I/O request inside a drain
section, you have to expect that more internal I/O might be going on
after the request has completed. If you don't want this, you have to
issue another bdrv_drain() or use a nested bdrv_drained_begin/end()
section.

Sounds reasonable enough to me, but I guess this should be explicitly
documented.

Kevin