Re: CFS flat runqueue proposal fixes/update

From: Rik van Riel
Date: Fri Aug 07 2020 - 15:42:06 EST

On Fri, 2020-08-07 at 16:14 +0200, Dietmar Eggemann wrote:
> On 31/07/2020 09:42, Rik van Riel wrote:
> > Possible solution
> > ...
> I imagine that I can see what you want to achieve here ;-)
> But it's hard since your v5 RFC
> is
> pretty old by now.

The v5 patches also do not implement this new idea, and
still suffer from the corner cases that Paul Turner
pointed out last year.

> Do you have a version of the patch-set against tip/sched/core? Quite
> a
> lot has changed (runnable load avg replaced by runnable avg, rq->load
> is
> gone, CFS load balance rework).

Not yet. We got a baby this spring, so I've been busy
with things like milk and diapers, instead of with

I wanted to get this proposal out before Plumbers, so
we could at least talk about it, and maybe find flaws
with this idea before I spend weeks/months implementing it :)

> IMHO it would be extremely helpful to have a current patch-set to
> discuss how these other problems can be covered by patches on top.

Agreed. I hope to get some time to work on that, but
no guarantees about getting it ready before Plumbers :)

All Rights Reversed.

Attachment: signature.asc
Description: This is a digitally signed message part