Re: EEVDF/vhost regression (bisected to 86bfbb7ce4f6 sched/fair: Add lag based placement)

[Peter Zijlstra]

On Wed, May 01, 2024 at 11:31:02AM -0400, Michael S. Tsirkin wrote:
> On Wed, May 01, 2024 at 12:51:51PM +0200, Peter Zijlstra wrote:

> > I'm still wondering why exactly it is imperative for t2 to preempt t1.
> > Is there some unexpressed serialization / spin-waiting ?
> 
> 
> I am not sure but I think the point is that t2 is a kworker. It is
> much cheaper to run it right now when we are already in the kernel
> than return to userspace, let it run for a bit then interrupt it
> and then run t2.
> Right, Tobias?

So that is fundamentally a consequence of using a kworker.

So I tried to have a quick peek at vhost to figure out why you're using
kworkers... but no luck :/

Also, when I look at drivers/vhost/ it seems to implement it's own
worker and not use normal workqueues or even kthread_worker. Did we
really need yet another copy of all that?

Anyway, I tried to have a quick look at the code, but I can't seem to
get a handle on what and why it's doing things.