[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Is: PVH - how to solve maxmem != memory scenario? Was:Re: [PATCH] libxl: create PVH guests with max memory assigned

On Tue, 5 Aug 2014 10:18:44 -0400
Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx> wrote:

> On Tue, Aug 05, 2014 at 01:08:22PM +0200, Roger Pau Monnà wrote:
> > On 05/08/14 11:34, David Vrabel wrote:
> > > On 05/08/14 09:55, Ian Campbell wrote:
> > >> On Thu, 2014-07-17 at 13:02 +0200, Roger Pau Monne wrote:
> > >>
> > >> Sorry for the delay replying, this somehow slipped through my


> It is evolving :-)
> > 
> > I've always thought of PVH as PVHVM without a device model, so IMHO
> > it would make more sense to use PoD rather than the PV ballooning
> > approach, but I would like to hear opinions from others before
> > taking a stab into implementing it.
> Lets rope Mukesh, Tim, George and Jan in here.
> Mukesh's feeling was that it is an PV.
> I believe George is the opinion of 'HVM' without the device model.
> In the past I  was thinking that since it is from the PV it would
> be more of that (PV) without the P2M and M2P. And the memory
> management (so E820) would follow the PV paths and do the proper
> ballooning/decreasing.
> However I think it was you (David) who suggested that we just
> setup the E820 properly in the toolstack/hypervisor and have it
> match the hypervisors' P2M. That I believe is what Roger's patch
> was aiming at.
> A bit of past history:
> Mukesh's initial patches (v3, see
> https://lkml.org/lkml/2012/10/17/553, and
> https://lkml.org/lkml/2013/12/12/627, for new hypercall) took the
> path that the PV guest will act as PV. And it will do the proper
> hypercalls to expand/contract the Xen's P2M to balloon out and in.
> However the only reason for this was to match the P2M (assuming it
> was flat and up to nr_pages) to E820 (which would be discontingous)
> and setup the correct EPT entries in the hypervisor.
> My personal opinion is that the easiest path is the best.

Agree. My opinion, that I had expressed to Roger few weeks ago when we
initially talked of which approach to take here, is we should pick
the best path in terms of performance and simplicity. My intention always 
has been to examine both PV and HVM, and pick the best, or if a third
option makes sense, go that way - that IMO is the whole point of PVH.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.