On Tue, Jul 10, 2018 at 11:26:45AM +0100, Ian Jackson wrote:
> Wei Liu writes ("Re: [PATCH] hvm/altp2m: Clarify the proper way to extend the
> altp2m interface"):
> > On Tue, Jul 10, 2018 at 10:33:22AM +0100, George Dunlap wrote:
> > > + * Normally hypercalls made by a program in domain 0 in order to
> > > + * control a guest would be DOMCTLs rather than HVMOPs. But in order
> > > + * to properly enable the 'internal' use case, as well as to avoid
> > > + * fragmentation, all altp2m subops should come under this single
> > > + * HVMOP.
> >
> > I don't understand this argument. There is no risk of code duplication /
> > fragmentation if the implementation is contained within a function.
> > Should we choose to split one HVMOP into a DOMCTL and a HVMOP, there is
> > now two entries to the internal function, each of which with proper
> > checks, but they will call the same internal function eventually.
> >
> > I admit I haven't followed the discussion closely.
>
> "each of which with proper checks". You end up doing some of the
> parameter validation twice. That is very undesirable indeed.
Fair enough.
Wei.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
