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

Re: [Xen-devel] [PATCH 1/8] viridian: add init hooks



> -----Original Message-----
> From: Xen-devel [mailto:xen-devel-bounces@xxxxxxxxxxxxxxxxxxxx] On Behalf
> Of Paul Durrant
> Sent: 03 January 2019 09:08
> To: Andrew Cooper <Andrew.Cooper3@xxxxxxxxxx>; xen-
> devel@xxxxxxxxxxxxxxxxxxxx
> Cc: Wei Liu <wei.liu2@xxxxxxxxxx>; Jan Beulich <jbeulich@xxxxxxxx>; Roger
> Pau Monne <roger.pau@xxxxxxxxxx>
> Subject: Re: [Xen-devel] [PATCH 1/8] viridian: add init hooks
> 
> > -----Original Message-----
> > From: Andrew Cooper
> > Sent: 02 January 2019 17:37
> > To: Paul Durrant <Paul.Durrant@xxxxxxxxxx>; xen-
> devel@xxxxxxxxxxxxxxxxxxxx
> > Cc: Wei Liu <wei.liu2@xxxxxxxxxx>; Jan Beulich <jbeulich@xxxxxxxx>;
> Roger
> > Pau Monne <roger.pau@xxxxxxxxxx>
> > Subject: Re: [Xen-devel] [PATCH 1/8] viridian: add init hooks
> >
> > On 02/01/2019 16:08, Andrew Cooper wrote:
> > > On 20/12/2018 16:33, Paul Durrant wrote:
> > >> This patch adds domain and vcpu init hooks for viridian features. The
> > init
> > >> hooks do not yet do anything; they will be added to by subsequent
> > patches.
> > >>
> > >> Signed-off-by: Paul Durrant <paul.durrant@xxxxxxxxxx>
> > > Please can we start by fixing the current, broken, initialisation and
> > > destruction logic ?
> > >
> > > To get rid of DOMCTL_setmaxvcpus, we need the *_destroy() logic to be
> > > fully idempotent.  Also, viridian_domain_deinit() should not call into
> > > viridian_vcpu_deinit(), and viridian_vcpu_deinit() shouldn't be faking
> > > up a write to the assist page.
> > >
> > > AFAICT, the deinit path is all entirely pointless at the moment and
> can
> > > be deleted.
> >
> > Given that we are now going to be allocating non-trivial structures for
> > viridian domains, I'd like to float the idea of changing viridian
> > initialisation to be a domaincreate flag, rather than blindly assuming
> > that all HVM guests want it.
> 
> That would certainly be cleaner but sounds non-trivial.

Looking at this a little more... Viridian is an x86 specific thing, so an extra 
flag in xen_arch_domainconfig would seem most appropriate. This would then need 
to wired into the appropriate place(s) in libxl. I'll have a look at how much 
work this is likely to be.

  Paul

> 
> >
> > I think this is going to be necessary part of fixing CPUID policy
> > derivation in the longterm, because regenerating the policy when
> > HVM_PARAM_VIRIDIAN changes isn't viable.
> >
> 
> Indeed, but again it's 'longterm'.
> 
>   Paul
> 
> > ~Andrew
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@xxxxxxxxxxxxxxxxxxxx
> https://lists.xenproject.org/mailman/listinfo/xen-devel
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel

 


Rackspace

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