Re: [Xen-devel] [PATCH v4 09/14] xen/x86: split Dom0 build into PV and PVHv2

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 16.12.16 at 18:57, <roger.pau@xxxxxxxxxx> wrote:
> On Fri, Dec 16, 2016 at 09:17:01AM -0700, Jan Beulich wrote:
>> >>> On 16.12.16 at 15:45, <roger.pau@xxxxxxxxxx> wrote:
>> > On Fri, Dec 16, 2016 at 02:28:54PM +0000, Roger Pau Monne wrote:
>> >> On Fri, Dec 09, 2016 at 09:07:16AM -0700, Jan Beulich wrote:
>> >> > >>> On 30.11.16 at 17:49, <roger.pau@xxxxxxxxxx> wrote:
>> >> > > @@ -1655,6 +1653,28 @@ out:
>> >> > >      return rc;
>> >> > >  }
>> >> > >  
>> >> > > +static int __init construct_dom0_hvm(struct domain *d, const 
>> >> > > module_t *image,
>> >> > > +                                     unsigned long image_headroom,
>> >> > > +                                     module_t *initrd,
>> >> > > +                                     void *(*bootstrap_map)(const 
>> >> > > module_t *),
>> >> > > +                                     char *cmdline)
>> >> > > +{
>> >> > > +
>> >> > > +    printk("** Building a PVH Dom0 **\n");
>> >> > 
>> >> > Why again is it that you call the function "hvm" but mean "pvh"?
>> >> 
>> >> This was to differentiate between the current "pvh" functions in this 
>> >> file that
>> >> refer to PVHv1. I could name them pvhv2, but IMHO hvm seems clearer and
>> >> shorter.
>> > 
>> > Oh, and the other reason was that Xen doesn't really know the difference
>> > between a HVM guest and a PVHv2 guest, hence hvm felt more natural.
>> 
>> Xen certainly can tell the difference for Dom0, since a true HVM
>> Dom0 can't exist without a lot of work towards getting a device
>> model run somewhere to service it. I continue to think that "hvm"
>> in any of the names involved in this series is misleading.
> 
> Yes, but that device model isn't a Xen-kernel component, and would be attached
> using the ioreq server machinery anyway, so IMHO I still think this is more
> like HVM rather than anything else from a Xen PoV. And if we ever got 
> something
> like a complete HVM Dom0 with a device model it would certainly use this
> machinery.
> 
> In any case, I don't want to start a bikeshed over it, so please tell me your
> preference for either pvh_ or pvh2 or pvhv2_ prefix for these new functions.

pvh please.

Jan



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