[Qemu-devel] Default for phys-addr-bits? (was Re: [PATCH 4/5] x86: Allow physical address bits to be set)

[Eduardo Habkost]

On Fri, Jun 17, 2016 at 12:49:05PM -0300, Eduardo Habkost wrote:
> On Fri, Jun 17, 2016 at 05:28:03PM +0200, Paolo Bonzini wrote:
> > On 17/06/2016 17:19, Eduardo Habkost wrote:
> > > > > So, what happens if the guest play tricks in bits 40-45 when QEMU
> > > > > sets the limit to 40 but we are running in a 46-bit host? Is it
> > > > > really a problem? I assumed it would be safe.
> > > > 
> > > > The guest expects a "reserved bit set" page fault, but doesn't get one.
> > > 
> > > Wait, are you talking about migration only, or are you really
> > > talking about running current QEMU (hardcoded to 40) on a 46-bit
> > > host? I'm not talking about migration, above.
> > 
> > I'm talking about both. :(
> > 
> > > We really can't emulate a 40-bit machine in a 46-bit host? I
> > > didn't expect that.
> > 
> > Unfortunately that's the case. :(
> 
> OK, so please ignore all my suggestions about choosing reasonable
> defaults based on VM size, etc.
> 
> Now all options look bad, and we need to choose the least harmful
> as the default... I need to re-read this thread to be able to
> give an opinion.
> 

I noticed we didn't continue to discuss what would be a good
default.

We can delegate this decision to libvirt, but we would still have
to evaluate the options and tell the libvirt folks what's the
default we recommend.

In an offline discussion, Paolo suggsted making host
phys-addr-bits the default, but I am not sure this is really the
best option.

The consequences of migrating (or having migration blocked) to a
host with smaller phys-addr-bits sound worse to me than the
consequences of just having guest's phys-addr-bits smaller than
the host's.

-- 
Eduardo