From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from [140.186.70.92] (port=47662 helo=eggs.gnu.org)
	by lists.gnu.org with esmtp (Exim 4.43) id 1Q2DrN-0005M7-49
	for qemu-devel@nongnu.org; Tue, 22 Mar 2011 22:31:46 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <anthony@codemonkey.ws>) id 1Q2DrM-0008Mx-1X
	for qemu-devel@nongnu.org; Tue, 22 Mar 2011 22:31:44 -0400
Received: from mail-yw0-f45.google.com ([209.85.213.45]:41864)
	by eggs.gnu.org with esmtp (Exim 4.71)
	(envelope-from <anthony@codemonkey.ws>) id 1Q2DrL-0008Mm-Tv
	for qemu-devel@nongnu.org; Tue, 22 Mar 2011 22:31:43 -0400
Received: by ywl41 with SMTP id 41so3837938ywl.4
	for <qemu-devel@nongnu.org>; Tue, 22 Mar 2011 19:31:43 -0700 (PDT)
Message-ID: <4D895B88.9080904@codemonkey.ws>
Date: Tue, 22 Mar 2011 21:31:36 -0500
From: Anthony Liguori <anthony@codemonkey.ws>
MIME-Version: 1.0
Subject: Re: [Qemu-devel] [PATCH 11/11] test-vmstate: add test case to verify
	we don't change VMState
References: <1300839376-22520-1-git-send-email-aliguori@us.ibm.com>
	<1300839376-22520-12-git-send-email-aliguori@us.ibm.com>
In-Reply-To: <1300839376-22520-12-git-send-email-aliguori@us.ibm.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
List-Id: qemu-devel.nongnu.org
List-Unsubscribe: <http://lists.nongnu.org/mailman/listinfo/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: <http://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
Cc: Jan Kiszka <jan.kiszka@siemens.com>, qemu-devel@nongnu.org, Juan Quintela <quintela@redhat.com>

On 03/22/2011 07:16 PM, Anthony Liguori wrote:
> We've had lots of issues surrounding live migration breaking.  This is because
> we haven't had a good way to validate that what we're migrating isn't changing
> underneath of us.
>
> This test works by first converting the in-tree schema to C source as a string.
> This is built into the test case.  The test case will then query QEMU for the
> current schema.
>
> It converts both strings to QObjects via the JSON parser and then does a
> recursive diff.  Unlike a simple diff command, this lets us say exactly what you
> did to break migration.  For instance, you get error messages like:
>
>    You broke migration by changing the type of field 'par' in device 'cpu',
>    version 1 from 'uint32' to 'uint64'.
>
> It only covers devices that support VMState and it currently doesn't look at
> subsections yet.
>
> The in-tree schema needs to be updated whenever migration changes but this ends
> up being a very nice way to ensure that we're reviewing migration protocol
> changes.
>
> The test case is built in the default build, and can be run by `make check'.

This test takes a disturbingly long time to run considering what it's 
doing (~3 seconds).  Turns out, the JSON parser is horribly slow because 
it allocates and frees memory faster than a hummingbird on steroids.

I have a local patch that removes this and afterwards the test runs in 
just 70ms.  I'll send that fix out tomorrow.

Regards,

Anthony Liguori