On Thu, 17 May 2012 12:33:55 +0930, Rusty Russell <rusty.russell at linaro.org> wrote: > > I was considering a more careful approach adding a new path in > > the interrupt vector in hyp mode, calling the switching code and > > returning to the guest, while avoiding going all the way to the host to > > do the switch. > > Right, that would be more efficient. We should probably just do that. I started implementing this, then realized it's a bit harder than that. If the host is doing lazy switching, the attempt to restore will trap. I'm not sure what'll happen at that point, but given we're still in hyp mode, it won't be good. Thus I think this fastpath would be an optimization which we could do IFF the FPEXC EN bit were set (ie. we're not lazy saving). We'll still need a full exit when the host kernel wants to restore the VFP state. Cheurs, Rusty.
