Re: [Patch v4 01/22] sched/cache: Introduce infrastructure for cache-aware load balancing

[Chen, Yu C]

On 4/10/2026 7:00 AM, Peter Zijlstra wrote:
> On Thu, Apr 09, 2026 at 12:21:58PM -0700, Tim Chen wrote:
>
> > > > +	    get_nr_threads(p) <= 1) {
> > >
> > > And this, I just noticed this, why are we excluding tasks with one
> > > thread? The comment just states we are (doh), but utterly fails to
> > > explain why.
> >
> > The thought was that for a single thread, we will tend to place it
> > in same LLC as it was previously running with cache hot.  Incurring
> > the added cost of tracking its residency in LLC is uneeded.
>
> Not sure -- the moment it gets migrated away, it will stay away. There
> is no force to pull it back, while with cache aware it will try and come
> back for a short while -- until it's build an occupancy on the new llc.

Besides Tim's scenario, in the original design, I thought the 
beneficiary of
cache-aware scheduling was mainly multi-threaded processes, where threads
share data with each other. Single-threaded processes normally exhibit
no data sharing. This check was also introduced to prevent a single process
 from concentrating excessively on one LLC. Nevertheless, we can omit this
check if the user provides hints for such single-threaded workloads, either
via prctl or other interfaces in the future.

> Anyway, if you want to do this, it needs to be a separate patch.

OK, we will move this check into a separate one.

thanks,
Chenyu