Proposal: Remove or reduce math-finite.h

[Wilco Dijkstra]

Hi Steve,

> I ran build-many-glibcs.py to build all of the glibs that are defined
> to build by that script and excluding csky which failed to build (PR
> 24371), here are the places where the pow, powf, log, logf, log2,
> log2f, exp, expf, exp2, and exp2f are not aliases of the _finite
> version of the matching routine.
>
> i486/i586/i686
>        On these platforms pow, log, log2, exp, and exp2 are not
>        aliases of the finite versions.

These are old x87 assembler implementations. Performance would
improve if they were removed.

> x86_64
>        On these platforms pow_finite, log_finite, and exp_finite are
>        IFUNCs but the non-finite versions of those routines are
>        regular functions.

I think that's unintentional given these use the generic functions.

> m68k
>        On this platform none of the functions are aliases.

It's not obvious what is best here. I don't know why anyone would care
about FP performance on 68k, most didn't have floating point hardware.

> powerpc
>        On this platform ONLY expf and __expf_finite are not aliases.

That's using an old assembler implementation, the generic implementation
will be faster.

Cheers,
Wilco