Re: [PATCH v3 3/3] tools: introduce parameter max_wp_ram_ranges.

[Paul Durrant <Paul.Durrant@citrix.com>]

> -----Original Message-----
> From: Jan Beulich [mailto:JBeulich@suse.com]
> Sent: 04 February 2016 14:13
> To: Paul Durrant
> Cc: Andrew Cooper; George Dunlap; Ian Campbell; Ian Jackson; Stefano
> Stabellini; Wei Liu; Kevin Tian; zhiyuan.lv@intel.com; Zhang Yu; xen-
> devel@lists.xen.org; Keir (Xen.org)
> Subject: RE: [Xen-devel] [PATCH v3 3/3] tools: introduce parameter
> max_wp_ram_ranges.
> 
> >>> On 04.02.16 at 14:47, <Paul.Durrant@citrix.com> wrote:
> >> From: Ian Jackson [mailto:Ian.Jackson@eu.citrix.com]
> >> Sent: 04 February 2016 13:34
> >>  * Is it possible for libxl to somehow tell from the rest of the
> >>    configuration that this larger limit should be applied ?
> >>
> >>    AFAICT there is nothing in libxl directly involving vgpu.  How can
> >>    libxl be used to create a guest with vgpu enabled ?  I had thought
> >>    that this was done merely with the existing PCI passthrough
> >>    configuration, but it now seems that somehow a second device model
> >>    would have to be started.  libxl doesn't have code to do that.
> >>
> >
> > AIUI if the setting of the increased limit is tied to provisioning a gvt-g
> > instance for a VM then I don't there needs to be extra information in the
> VM
> > config. These seems like the most sensible thing to do.
> 
> I don't understand this: For one, it's still unclear to me on what basis
> it would be known that a given VM is a "gvt-g instance". And even if
> that's indeed derivable from something, the uncertainty about a
> workable upper bound on the number of WP ranges would still seem
> to demand the value to be specifiable separately...

There are patches in the XenGT xen repo which add extra parameters into the VM config to allow libxl to provision a gvt-g instance (of which there are a finite number per GPU) for a VM. The increased limit could be applied when doing so and it may be feasible to determine (maybe from the version of the GPU h/w) what a reasonable limit is.

  Paul

> 
> >> I now understand that these mmio ranges are created by the device
> >> model.  Of course the device model needs to be able to create mmio
> >> ranges for the guest.  And since they consume hypervisor resources,
> >> the number of these must be limited (device models not necessarily
> >> being trusted).
> >
> > ...but I think there is still an open question as to whether the toolstack
> > is allowed to set that limit for a VM or not. IMO the toolstack should be
> > allowed to set that limit when creating a domain.
> 
> ... as you indicate here.
> 
> Jan