Re: [Xen-devel] [RFC PATCH 1/8]: PVH: Basic and preparatory changes

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Thu, 2012-09-13 at 18:36 +0100, Pasi Kärkkäinen wrote:
> On Thu, Sep 13, 2012 at 10:14:39AM -0700, Mukesh Rathor wrote:
> > On Thu, 13 Sep 2012 07:00:07 +0100
> > Ian Campbell <Ian.Campbell@xxxxxxxxxx> wrote:
> > > The toolstack (or the dom0 builder) can make the determination whether
> > > to enable this mode itself itself based on the notes (although it is
> > > always handy to have an override to force things for debug).
> > 
> > Hmmm... So you are suggesting that if hardware and xen supports it, then
> > PV should just boot in PVH mode and user has no way to override it. I
> > don't know if that's a good idea in the short term, perhaps in the long
> > term. Let me discuss with folks here a bit.
> > 
> 
> I read that as "if hardware, xen and kernel supports PVH, enable it as a 
> default,
> but still allow the user to force enable/disable it in the domain cfgfile for 
> easier debugging".

That's right. I have no objection to an override, in fact I think it's a
requirement to be able to force this sort of feature on/off (to aid for
debugging, account for differing workload requirements and allowing the
test system to test both configurations etc).

Long term the aim should be for the default to be on whenever possible.
Maybe the default should remain off for the time being, or maybe we want
to get as much exposure of PVH as possible ASAP, which would argue for
it being on by default in unstable, perhaps with a decision being made
around freeze time as to the default in 4.3.

The feature declarations in the ELF note also allow the toolstack to
sanity check what the user asked for and report an error if it isn't
possible, rather than booting a kernel in a mode which cannot possibly
work.

Ian.

> 
> -- Pasi
> 



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