Re: [Qemu-devel] [PATCH 00/10] RFCv3: vhost-user: simple reconnection support

[Michael S. Tsirkin]

On Fri, May 13, 2016 at 08:30:36PM +0200, Marc-André Lureau wrote:
> Hi
> 
> On Tue, May 10, 2016 at 6:28 PM, Marc-André Lureau <address@hidden> wrote:
> > Hi
> >
> > ----- Original Message -----
> >> On Tue, May 10, 2016 at 06:03:50PM +0200, address@hidden wrote:
> >> > From: Marc-André Lureau <address@hidden>
> >> >
> >> > Hi,
> >> >
> >> > In a previous series "RFCv2: vhost-user: shutdown and reconnection", I
> >> > proposed to add a new slave request to handle graceful shutdown, for
> >> > both qemu configuration, server or client, while keeping the guest
> >> > running with link down status.
> >>
> >> OK so I would say patches 1-4 are bugfixes, looks like they
> >> can be Cc stable even?
> >
> > 4 is being used by 5 and 10.
> > 2-3 are only for testing.
> >
> > 4-8 are nice to have as they avoid obvious problems/crashes when handling 
> > disconnected state and add basic reconnection checks.
> >
> > 9 was already considered for stable by Eric in a previous series
> >
> > 10 would be good to have if 1 is accepted, to check the minimum works as 
> > expected
> >
> 
> FYI, I have a follow up series (~20 patches,
> https://github.com/elmarco/qemu/tree/vhost-user-reconnect) doing
> mostly cleanups and extra checks for disconnection at run time. In
> particular, it should avoid some obvious crashers/asserts, and
> prevents qemu from running as long the initial vhost_user_start()
> didn't succeed (so initial flags are set). I would like to know how to
> proceed with the follow-up: should I resend the whole series or should
> we review/merge this rfc first (even though it is known to be
> incomplete in many disconnect cases that the follow up fixes).
> 
> thanks

I think a gradual merge is better.

> -- 
> Marc-André Lureau