Re: [Xen-devel] QNX Neutrino and RT-Xen

[Dario Faggioli <dario.faggioli@xxxxxxxxxx>]

On ven, 2014-01-17 at 16:59 +0000, Ian Campbell wrote:
> On Fri, 2014-01-17 at 18:50 +0200, Pavlo Suikov wrote:
> 
> > does anyone know about efforts on bringing QNX Neutrino as a dom0 or
> > domU in Xen - or, more specifically, in RT-Xen with
> > global/partitioning schedulers which should potentially support
> > real-time requirements for target OS?
> > 
> > 
> > Existing papers on RT-Xen imply using Linux as a target system, and
> > all QNX mentions with regard to Xen are quite outdated. I see that
> > RT-Xen activities are quite recent and production applications may
> > still be absent, but maybe anyone has tried this yet?
> 
> I don't know much (or anything) about RT-Xen but from a regular Xen PoV
> it's been a long while since I've heard anything about QNX on x86 Xen,
> and I've never heard anything about QNX on ARM Xen (at least the h/w
> assisted port in mainline).
> 
Well, it's been a while since the last time I tried QNX, and I *never*
used is on Xen (only baremetal). In principle, but on x86, it should not
be a big deal to have it running inside an HVM domain.

Of course, even there, I don't expect you to be able to get advantage of
any of the PV, nor PVHVM stuff, that are available in Linux (e.g., in
PVHVM mode) or even in Windows (with the proper drivers installed)
guests... Unless some kind of PV drivers exist for QNX, but I'm not
aware of any.

Getting reasonable RT performances is another pair of hands... We'll
(well, at list I will) be looking at that in the very immediate future,
so stay in touch. :-P

For ARM, given the fact that there is no a real HVM mode there, I think
you'd need a proper port of QNX on Xen on ARM, in its own and full
right.

This is probably going to be a lot similar to what is being attempted
here:
http://bugs.xenproject.org/xen/mid/%3C1387278345.8738.80.camel@Solace%3E

> If you are interested specifically in RT-Xen then I think you will
> probably have more luck on their mailing list, not much (essentially no)
> RT-Xen stuff happens on this list.
> 
Indeed you should ask them, and I'm adding Sisu to the recipients of
this e-mail.

That being said, RT-Xen is "just" two new schedulers (each with a couple
of possible 'operational mode', though), so porting on RT-Xen rather
than porting on Xen should not make that much of a difference.

Again, if thee will be something missing on the side of enabling
real-time behavior in the guest, that's a separate issue, and it is,
again, something that applies to both the cases (RT-Xen and Xen) pretty
much in the same way.

Anyway, again, there is rising interest in this sort of workloads these
days, as Artem (from your same company, I think, and I'm adding him, and
a bunch of other people too, to the Cc list) knows. :-)

In summary, I don't think anyone has ever tried that, given the above,
if you decide to do that, feel free to ask for any kind of help and keep
us posted on how it's going...  :-)

Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

Attachment: signature.asc
Description: This is a digitally signed message part

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