On Mon, 2012-09-03 at 18:59 +0300, Avi Kivity wrote: > On 08/29/2012 11:49 AM, Peter Maydell wrote: > > On 29 August 2012 09:47, Jan Kiszka <jan.kis...@siemens.com> wrote: > >> On 2012-08-28 23:26, Peter Maydell wrote: > >>> Since this is arch-specific we should probably give the > >>> resulting device a more specific name than "pci-assign", > >>> which implies that it is (a) ok for any PCI system and > >>> (b) not something that will be obsolete in the foreseen > >>> future. > >> > >> The name has to be like this for libvirt compatibility. I can provide it > >> as alias, though, calling the device "kvm-pci-assign" by default. Better? > > > > ...is it likely to ever work for non-x86 PCI devices? > > It used to work for ia64 before it died. So it's not architecture > specific by design.
Caveat; ia64 is a different architecture, but it was on Intel VT-d chipsets, which therefore supported the IOMMU API and ia64 maps guests the same way. This is the same problem I ran into trying to name the vfio iommu driver for x86. It's not really x86-specific, but it's got to look and smell pretty similar. Thanks, Alex -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
