Re: [PATCH v11 3/4] softmmu/dirtylimit: implement virtual CPU throttle

[Hyman Huang]

在 2022/1/20 18:56, Peter Xu 写道:
> On Thu, Jan 20, 2022 at 06:39:01PM +0800, Hyman Huang wrote:
> > > It only runs once per second so it shouldn't bring a lot of differences, but
> > > I'm also wondering whether we should also turn that temporarily off too when
> > > dirtylimit is enabled - we can simply let it keep sleeping if dirtylimit is in
> > > service.
> > Does this work ok when dirtylimit and migration happens concurrently?
> > Migration may fetch the aged dirty bitmap info from slot if we turn reaper
> > thread off. As you metioned above, reaper thread only runs once per second.
> > Is it more suitable for not touching the reaper thread logic?
>
> Yes I think it'll still work, as migration will do explicit sync/collect after
> each iteration.  See kvm_dirty_ring_flush().
Oh, i get it. :)
I just missed the key function kvm_dirty_ring_flush in kvm_dirty_ring_flush.
> In short, you shouldn't touch the rest of the kvm_dirty_ring_reap() callers,
> they'll always flush all the rings.
>
> But it'll be nicer if you will try it out. :)
Ok, i'll try this out.

--
Best regard

Hyman Huang(黄勇)