Re: [V3 PATCH 5/9] PVH dom0: implement XENMEM_add_to_physmap_range for x86

[Mukesh Rathor <mukesh.rathor@oracle.com>]

On Mon, 02 Dec 2013 12:47:25 +0000
"Jan Beulich" <JBeulich@suse.com> wrote:

> >>> On 27.11.13 at 03:27, Mukesh Rathor <mukesh.rathor@oracle.com>
> >>> wrote:
> > This preparatory patch adds support for XENMEM_add_to_physmap_range
> > on x86 so it can be used to create a guest on PVH dom0. To this
> > end, we add a new function xenmem_add_to_physmap_range(), and change
> > xenmem_add_to_physmap_once parameters so it can be called from
> > xenmem_add_to_physmap_range.
> > 
> > Please note, compat will continue to return -ENOSYS.
> 
> And as noted a number of times before - I don't think that's
> appropriate. There's nothing keeping non-PVH guests from using
> this interface, and hence it should either be uniformly available to
> all of them, or uniformly unavailable. I'm not intending to apply
> such a half baked thing.

I am sorry for your frustration, but as I said before, doing it is
not straightforward as it involves creating a new version of 
XLAT_add_to_physmap which unfortunately is generated via python/shell
scripts. At this point, I want to focus on the 64bit. I understand while
it's not a regression, it's a partial implementation, but so is the entire
PVH and many other features that are "work in progress". Your final word
on this was:

quote:
Then leave it out, and I'll waste my time on getting it implemented
once the patch set is in. But please add a clear note of this state
to the patch description.

Jan
end quote

All your other comments are taken care of. Thanks for your help.

mukesh