qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH] Migration compatibility for serial


From: Dr. David Alan Gilbert
Subject: Re: [Qemu-devel] [PATCH] Migration compatibility for serial
Date: Wed, 17 Jun 2015 10:38:30 +0100
User-agent: Mutt/1.5.23 (2014-03-12)

* Paolo Bonzini (address@hidden) wrote:
> 
> 
> On 17/06/2015 10:37, Dr. David Alan Gilbert wrote:
> >> > On 16/06/2015 20:54, Dr. David Alan Gilbert (git) wrote:
> >> > > From: "Dr. David Alan Gilbert" <address@hidden>
> >> > > 
> >> > > Older QEMUs dont understand the new (sub)sections that
> >> > > may be generated in the serial device.   Limit their generation
> >> > > to newer machine types.
> >> > > 
> >> > > Signed-off-by: Dr. David Alan Gilbert <address@hidden>
> > > 
> > > No, please.  Upstream QEMU doesn't want to get into judgement about when
> > > migration quality might be "good enough" that you can drop subsections.
> > 
> > Other people disagree with that statement.
> > Who upstream doesn't want it?
> 
> That's always been the policy as far as I know.  Certainly I don't.
> When we were working on RHEL7, there were quite a few discussions about
> this; I remember Orit Wassermann also was a proponent of the "clean
> slate" approach.

> The problem is that if you want bug compatibility, you also want a point
> (e.g. a major release) where you can start from a clean slate and drop
> all compatibility hacks.  Upstream there is no such point.

It doesn't necessarily have to be a clean slate; the other way to do it is
to have a version cut off, so you don't have any bug-compatibility
prior to a particular version, and then roll that cut off point forward,
much in the same way we do for glib version dependencies.
My flag naming of 'serial_migrate_pre_2_2' at least made it clear
where the line was for that feature.

> It's already hard enough to ensure compatibility of versioned machine
> types, which are static, up to QEMU 0.10 or so; imagine what it would be
> like to guarantee the same for migration six or seven years down the
> line, considering how extremely data-driven migration is.

I agree it's not easy, and indeed this serial fix is just one of a whole
bunch of fixes that are needed.  But if you never try then you never
get any compatibility.

Dave

> 
> Paolo
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK



reply via email to

[Prev in Thread] Current Thread [Next in Thread]