On Mon, 2005-01-24 at 17:31 -0600, David Boutcher wrote: > virtualization-bounces@xxxxxxxxxxxxxx wrote on 01/24/2005 02:01:31 PM: > > If anybody here has virtual bus code that they would like > > to see reused in Xen, please follow up to this mail so we > > can get things started soon. > > I would hate to see a xen virtual bus...I would love to see > the power5 VIO bus move out of arch/ppc64 and into somewhere > generic. Well, "xen virtual bus" meaning here "a virtual bus being used by xen". VIO isn't insane to think about (in fact, I've looked quite a bit at it since I'm familiar with the power5 stuff :), but as it stands right now, the VIO stuff is fairly ugly (imho) since the entire file is basically #ifdef PSERIES ... #endif #ifdef ISERIES ... #endif And if we're just going to end up adding another #ifdef SOMETHING_ELSE, then why try to pretend they're the same thing? > I'm pretty UML ignorant...but is there any overlap with > the way UML does devices? My vague recollection of UML was that it presented just PCI devices. This seems like fairly overkill with Xen... Jeremy