On Mon, Apr 01, 2019 at 04:09:32PM +0200, David Hildenbrand wrote: > > > > When you say yield, I would guess that would involve config space access > > to the balloon to flush out outstanding hints? > > I rather meant yield your CPU to the hypervisor, so it can process > hinting requests faster (like waiting for a spinlock). This is the > simple case. More involved approaches might somehow indicate to the > hypervisor to not process queued requests but simply return them to the > guest so the guest can add the isolated pages to the buddy. If this is > "config space access to the balloon to flush out outstanding hints" then > yes, something like that might be a good idea if it doesn't harm > performance. The problem would be in testing this unfortunately. Same as any OOM hack, it is difficult to test well. > -- > > Thanks, > > David / dhildenb
