On 05/05/2015 03:22 AM, Ján Tomko wrote:
On Mon, May 04, 2015 at 01:15:45PM -0400, Laine Stump wrote:
> On 05/04/2015 08:59 AM, Ján Tomko wrote:
> Perhaps a reasonable compromise would be this:
>
> 1) rename the s390-piix-controllers test to
> "s390-allow-bogus-usb-controller" (or something like that) and remove
> the ide controller from it.
>
> 2) likewise rename "s390-usb-none" to
"s390-allow-bogus-usb-controller-none"
>
> (the idea of both of these is to make it apparent to the untrained eye
> that this is testing for something that isn't really valid, but we're
> allowing for now).
>
> 3) in a separate patch, add a VIR_WARN() any time a domain has a USB
> controller but doesn't support USB.
>
> 4) (already done in an upcoming patch) log an error if a domain has an
> ide controller defined when it isn't supported.
>
> How much (if any :-) or that would you agree to?
>
All of it seems reasonable to me.
I posted a new version of this patch that does items (1) and (2).
https://www.redhat.com/archives/libvir-list/2015-May/msg00134.html
The patch just after that on the list does (4). I didn't have the energy
to find the best place to do (3) today; maybe later.