On 16/08/21 16:03, Srikar Dronamraju wrote:

quoted

Your version is much much better than mine.
And I have verified that it works as expected.

Hey Peter/Valentin

Are we waiting for any more feedback/testing for this?

I'm not overly fond of that last one, but AFAICT the only alternative is
doing a full-fledged NUMA topology rebuild on new-node onlining (i.e. make
calling sched_init_numa() more than once work). It's a lot more work for a
very particular usecase.

--
Thanks and Regards
Srikar Dronamraju