Re: [Xen-devel] [PATCH 00 of 11 v5] NUMA aware credit scheduling

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Tue, 2013-04-16 at 16:22 +0100, George Dunlap wrote:
> On Tue, Apr 16, 2013 at 2:14 PM, Jan Beulich <JBeulich@xxxxxxxx> wrote:
> >>>> On 16.04.13 at 14:16, George Dunlap <George.Dunlap@xxxxxxxxxxxxx> wrote:
> >> On Tue, Apr 16, 2013 at 9:44 AM, Dario Faggioli
> >> <dario.faggioli@xxxxxxxxxx> wrote:
> >>> On gio, 2013-04-11 at 13:41 +0100, George Dunlap wrote:
> >>>> Keir,
> >>>>
> >>> Ping?
> >>
> >> Given the state we are in the release cycle, and the importance of
> >> this patch, and the relative simplicity of the changes in Xen outside
> >> of the scheduler:
> >>
> >> Tim and Jan, would one of you be willing to use an "Apply unless Keir
> >> objects" policy?
> >
> > Going through patches 1...5 again,
> >
> > - 1 and 2 touch tools/, but don't have a tools maintainer ack yet
> > - 1 touches the public interface (even if only the one affecting
> >   the tools)
> ...
> > - 5 haQs a public interface change again
> 
> I thought the policy was that the tools & the hypervisor have to
> remain in lockstep -- only libxl and the kernel interfaces need to be
> backwards-compatible, right?  Do either of these changes affect those?

AFAICT they are all domctls which are "private" interfaces between the
tools (libxc) and the hypervisor (i.e. one of the bits which needs to be
lockstep and is not considered a stable interface).

There's a lot more scope for fixing up errors in the domctl interface
than with a regular hypercall but it's still a hypercall interface and I
can appreciate why Jan would be wary of over stepping his boundaries WRT
Acking changes to those.

Ian.



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel