Re: [PATCH RFCv3 6/9] s390x/diag: subcode to query device memory region

[Cornelia Huck]

On Mon, 27 Jul 2020 14:02:47 +0200
David Hildenbrand <david@redhat.com> wrote:

> On 27.07.20 13:15, Heiko Carstens wrote:
> > On Mon, Jul 27, 2020 at 12:12:02PM +0200, David Hildenbrand wrote:  
> >>>>>> +#define DIAG500_DEVICE_MEMORY_REGION   4    
> >>>>>
> >>>>> Regardless what we end up with, this needs to be specified
> >>>>> somewhere(tm).  
> >>>>
> >>>> Yeah, there, we should also document the existing subcodes. What would
> >>>> be the right place for this? The kernel feels somewhat wrong to me.  
> >>>
> >>> The still supported subcode 3 is properly specified in the virtio spec.
> >>> That's not a good place for that new one, though.
> >>>
> >>> QEMU is probably a better place than the kernel to specify stuff,
> >>> although it's not really ideal, either. OTOH, do we ever expect other
> >>> hypervisors to implement this new subcode?  
> >>
> >> cloud-hypervisor implements virtio-mem. If it were ever to support s390x
> >> (guess it does not yet), it would also want to implement that one. But
> >> then, it can just look at QEMU doc I guess :)  
> > 
> > It must be well defined and easy to find also for kernel developers
> > who actually have to care about memory detection code :)  
> 
> So I'd suggest documenting it in QEMU (docs/specs ...) for now, and
> referencing it from the relevant Linux patch - other suggestions?

That's probably the easiest way for now... the kernel's s390-diag.rst
should also point to it.

However, I think we really need a central place for definitions that
are not just a Linux/QEMU interface, but can potentially also be used
by other hypervisors/guests. Nothing as complicated as an OASIS spec,
but maybe a git??b project?