Re: [PATCH v12] introduce XENMEM_reserved_device_memory_map

[Ian Campbell <ian.campbell@citrix.com>]

On Wed, 2015-07-22 at 07:14 -0600, Jan Beulich wrote:
> > 
> > > > On 22.07.15 at 14:52, <ian.campbell@citrix.com> wrote:
> > Looking through the older comments, it seems like there was a
> > conclusion to have an XSM check here, which isn't present here?
> 
> Hmm, did I lose track of such a request?

You asked Daniel in <548584C9020000780004DAB2@mail.emea.novell.com> and
there was a short thread until you said "Right, in that case we
definitely would need a check." in 
<5485E5DB020000780004DE0B@mail.emea.novell.com>.

That was from "[v8][PATCH 03/17] introduce
XENMEM_reserved_device_memory_map".

> > It looks like Tim and Ian's comments have been addressed (so far as
> > they were agreed to at the time).
> > 
> > WRT the comments Julien raised: at some point (early on) you said 
> > this
> > was only intended to be used by the toolstack. In which case can it 
> > not
> > be done in one of the unstable interfaces (e.g. sysctl I suppose is 
> > the
> > obvious one)?
> 
> Oh, indeed. There was so much back and forth here - I thought
> hvmloader is to use this interface (and it was doing so at some
> point), but that's not the case. So yes, we don't really need to
> guarantee stability (but we'd need to move the definitions into
> a suitable block then).

Ah, I hadn't realised that some of XENMEM was tools only. Yes this
should certainly be moved in there.

>  Still I think the altered layout with the
> union at the end is preferable (if only to reduce churn).

Agreed.

Ian.