Re: [Qemu-devel] [PATCH v1 3/3] s390x/sclp: extend SCLP event masks to 64 bits

[Cornelia Huck <cohuck@redhat.com>]

On Wed, 21 Feb 2018 17:51:19 +0100
Claudio Imbrenda <imbrenda@linux.vnet.ibm.com> wrote:

> On Wed, 21 Feb 2018 16:34:59 +0100
> Cornelia Huck <cohuck@redhat.com> wrote:
> 
> > On Tue, 20 Feb 2018 19:45:02 +0100
> > Claudio Imbrenda <imbrenda@linux.vnet.ibm.com> wrote:

> > > +static const VMStateDescription vmstate_event_facility_mask64 = {
> > > +    .name = "vmstate-event-facility/mask64",
> > > +    .version_id = 0,
> > > +    .minimum_version_id = 0,
> > > +    .needed = vmstate_event_facility_mask64_needed,
> > > +    .pre_load = vmstate_event_facility_mask64_pre_load,
> > > +    .fields = (VMStateField[]) {
> > > +        VMSTATE_UINT64(receive_mask, SCLPEventFacility),
> > > +        VMSTATE_END_OF_LIST()
> > > +     }
> > > +};
> > > +    
> > 
> > Are there plans for extending this beyond 64 bits? Would it make sense  
> 
> I don't know. I'm not even aware of anything above 32 bits, but since we
> are already using all of the first 32 bits, it's only matter of time I
> guess :)
> 
> > to use the maximum possible size for the mask here, so you don't need
> > to introduce yet another vmstate in the future? (If it's unlikely that  
> 
> That's true, but it requires changing simple scalars into bitmasks.
> Surely doable, but I wanted to touch as little as possible.

OK, that pushes this firmly into the 'overkill' area. Let's just go
with your current approach.

> 
> > the mask will ever move beyond 64 bit, that might be overkill, of
> > course.)