Re: [PATCH 16/27] KVM: Move kvm_guest_init out of generic code

On 02.07.2010, at 09:41, Geert Uytterhoeven wrote:

On Thu, 1 Jul 2010, Alexander Graf wrote:

quoted

Currently x86 is the only architecture that uses kvm_guest_init(). =

With

quoted

PowerPC we're getting a second user, but the signature is different =

there

quoted

and we don't need to export it, as it uses the normal kernel init =

framework.

=20
Making the signatures match (i.e. always return `int') wouldn't hurt,
since kvm_guest_init() apparently can fail on x86, too.

I'm reasonably indifferent here. Fact is that the x86 hook is done =
completely different from how we do it on ppc. So whatever we do, the =
signature doesn't belong in generic code.

If you like, feel free to send a follow-up patch making the x86 =
signature return failures :). I personally don't think it make sense to =
expose failures for PV speedups - they should never be mandatory and =
thus failure is no problem for the system, so the caller doesn't need to =
know.

Alex