From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:34093)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1Xnnjd-00009Q-U3
	for qemu-devel@nongnu.org; Mon, 10 Nov 2014 07:06:24 -0500
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1XnnjX-00071j-M1
	for qemu-devel@nongnu.org; Mon, 10 Nov 2014 07:06:17 -0500
Received: from mx1.redhat.com ([209.132.183.28]:56466)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <kwolf@redhat.com>) id 1XnnjX-00071d-EI
	for qemu-devel@nongnu.org; Mon, 10 Nov 2014 07:06:11 -0500
Date: Mon, 10 Nov 2014 13:06:04 +0100
From: Kevin Wolf <kwolf@redhat.com>
Message-ID: <20141110120604.GD4158@noname.str.redhat.com>
References: <4e27237d.8773.14989687829.Coremail.magazine.lihuiba@163.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <4e27237d.8773.14989687829.Coremail.magazine.lihuiba@163.com>
Subject: Re: [Qemu-devel] guest sync and bdrv_co_flush
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: lihuiba <magazine.lihuiba@163.com>
Cc: qemu-devel@nongnu.org

Am 07.11.2014 um 09:39 hat lihuiba geschrieben:
> Hi, all
> 
> I'm a user of qemu/kvm, and I'm wondering some internals of qemu/kvm, so I'd
> better post it in 
> this developer's mailing list.
> 
> To be specific, I'm wondering how data is flushed to disk. Intuitively, when
> the guest issues a 
> SYNCHRONIZE CACHE command in the SCSI layer, qemu/kvm should call bdrv_co_flush
> (),
> which will eventually call the block driver's bdrv_co_flush_to_os()
> and bdrv_co_flush_to_disk().
> 
> But from simple grep-ing, I didn't find any calling to bdrv_co_flush() is
> responsible for SYNCHRONIZE 
> CACHE. So, can you tell me how qemu/kvm ensures guest data be written on
> persistent storage?

The relevant code is in hw/scsi/scsi-disk.c, in the function
scsi_disk_emulate_command(), for 'case SYNCHRONIZE_CACHE:'.

The call chain from there is blk_aio_flush() -> bdrv_aio_flush() ->
bdrv_aio_flush_co_entry() -> bdrv_co_flush()

Kevin