Re: [PATCH 2/2] sched/fair: Scale wakeup granularity relative to nr_running

[Peter Zijlstra]

On Wed, Sep 22, 2021 at 02:20:02PM +0100, Mel Gorman wrote:
> 
> Note that the TuneD throughput-performance profile allows up to 15ms
> for sysctl_sched_latency (ignoring scaling) but there is no explanation
> why such a long period was necessary or why sched_latency_ns is also
> not adjusted. The intent may have been to disable wakeup preemption
> or it might be an oversight.  An internet search for instances where
> sysctl_sched_wakeup_granularity parameter are tuned to high values offer
> either no explanation or a broken one.

FWIW, if one wants to disable wakeup preemption, we've got SCHED_BATCH
for that.