Re: [RFC] Scheduler: DMA Engine regression because of sched/fair changes

[Mel Gorman]

On Tue, Jan 18, 2022 at 10:04:48AM +0800, Hillf Danton wrote:
> > > > 5) What DMA Engine enabled drivers (and dmatest) should use as design
> > > > pattern to conform migration/cache behavior? Does scheduler optimisation
> > > > conflict to DMA Engine performance in general?
> > > > 
> > > 
> > > I'm not familiar with DMA engine drivers but if they use wake_up
> > > interfaces then passing WF_SYNC or calling the wake_up_*_sync helpers
> > > may force the migration.
> > > 
> > 
> > Thanks for the advice. I'll try to check if this is a solution.
> 
> Check if cold cache provides some room for selecting CPU.
> 
> Only for thoughts now.
> 

That will still favour migrating tasks between CPUs that share LLC cache
at the expense of losing some higher level caches and cpufreq state
(depending on the CPUfreq governor).

-- 
Mel Gorman
SUSE Labs