On Tue, 14 Jun 2011, Jan Kiszka wrote: > I bet the motivation is to have an IRQ route that is independent of > QEMU, thus can be discovered inside the Xen kernel and then remains > stable - or is simply hard-wired. Device assignment? Direct legacy IRQ > injection from the kernel? >
This code predates me, so Keir might have more insights on the original motivation, but you are correct, they are hard-wired and can be used for device-assignment as well.