From mboxrd@z Thu Jan  1 00:00:00 1970
From: Stefan Hajnoczi <stefanha@gmail.com>
Subject: Re: AIO requests may be disordered by Qemu-kvm iothread with disk
 cache=writethrough, Bug or Feature?
Date: Wed, 14 Oct 2015 10:32:35 +0100
Message-ID: <20151014093235.GA14874@stefanha-thinkpad>
References: <dae53c1.1a208.15047527458.Coremail.charlie.song@ucloud.cn>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: kvm <kvm@vger.kernel.org>
To: "charlie.song" <charlie.song@ucloud.cn>
Return-path: <kvm-owner@vger.kernel.org>
Received: from mail-wi0-f179.google.com ([209.85.212.179]:37997 "EHLO
	mail-wi0-f179.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S932264AbbJNNvj (ORCPT <rfc822;kvm@vger.kernel.org>);
	Wed, 14 Oct 2015 09:51:39 -0400
Received: by wieq12 with SMTP id q12so84430883wie.1
        for <kvm@vger.kernel.org>; Wed, 14 Oct 2015 06:51:38 -0700 (PDT)
Content-Disposition: inline
In-Reply-To: <dae53c1.1a208.15047527458.Coremail.charlie.song@ucloud.cn>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On Thu, Oct 08, 2015 at 07:59:56PM +0800, charlie.song wrote:
>     We recently try to use Linux AIO from guest OS and find that the IOthread mechanism of Qemu-KVM will reorder I/O requests from guest OS 
> even when the AIO write requests are issued from a single thread in order. This does not happen on the host OS however.

I think you are describing a situation where a guest submits multiple
overlapping I/O requests at the same time.

virtio-blk does not guarantee a specific request ordering, so the
application needs to wait for request completion if ordering matters.

io_submit(2) also does not make guarantees about ordering.

Stefan