On Mon, Aug 12, 2024 at 10:04:07AM -0400, Waiman Long wrote: > > Anyway, using DIV_ROUND_UP() is a slight change in behavior as chunk_size > will be increased by 1 in most cases. I am a bit hesitant to make this > change without looking into more detail about the rationale behind the > current code. I don't think it matters much. Just look at the two lines after the division, they're both rounding the value up. So clearly this is expected to handle the case where work gets bunched up into the first N CPUs, potentially leaving some CPUs unused. But Daniel wrote the code so he can have the last say of whether we should round up after the division or after the other two ops. Daniel? Cheers, -- Email: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
