Re: [PATCH v2 0/5] lib/find: add find_nth_bit()

[Yury Norov]

On Mon, Jul 11, 2022 at 1:55 AM Andy Shevchenko
<andy.shevchenko@xxxxxxxxx> wrote:
>
> On Mon, Jul 11, 2022 at 6:51 AM Yury Norov <yury.norov@xxxxxxxxx> wrote:
> >
> > Kernel lacks for a function that searches for Nth bit in a bitmap.
> > Usually people do it like this:
> >         for_each_set_bit(bit, mask, size)
> >                 if (--n == 0)
> >                         return bit;
> >
> > Which is not so elegant, and very slow.
> >
> > This series adds fast routines for this task, and applies them where
> > appropriate.
> >
> > While here, move thin wrappers around find_bit() in nodemask.c to a
> > corresponding header, and because nodemask.c is empty after that,
> > remove it.
> >
> > v1: https://lore.kernel.org/lkml/20220706182300.70862-4-yury.norov@xxxxxxxxx/T/
> > v2: - count Nth bit from 0 (was 1);
> >     - use 'invert' trick in _find_nth_bit(), as in _find_next_bit();
> >     - cleanup comments;
> >     - fns() is kept inline - looking at __ffs() generic implementation,
> >       I decided to keep it untouched.
>
> Two observations:
> 1) your patches are not versioned (hint: use `git format-patch
> --thread -vX --cover-letter ...`, where X is a version you want to
> give);

OK

> 2) fns() is not good abbreviation, because among ffs (First) and fls
> (Last), fns would be read as Next, which is misleading, I'm not sure
> fnths(), which is correct, is good for readers.

I agree that fns() may be confusing, but fnths() is even worse to me.
I expect that it will be mostly used indirectly via find_nth_bit(), and
will not create a lot of confusion for users.

Thanks,
Yury