Add r-ccu support for the Allwinner a83t

[kevans]

Summary

The r-ccu on the a83t differs from the others only by what it names the ar100 parents. Export the _CCU macros so ccu_sun8i_r can differentiate between a83t r-ccu and the others, add the compat string for the a83t r-ccu.

NOTE: As far as I can tell, this is the only difference

[kevans]

Note: This patch is against -HEAD + D13205.

[manu]

sys/arm/allwinner/clkng/aw_ccung.h

33 ↗

(On Diff #35637)

I prefer that every consumer include this file instead of relying on header pollution.

sys/arm/allwinner/clkng/ccu_sun8i_r.c

130 ↗

(On Diff #35637)

I'd rather keep the array as all those register_clocks function are the same I'll merge them into one one day.

[kevans]

Unconditionally define the *_CCU macros in aw_ccung.h; maintain array-based clock approach in ccu_sun8i_r

[manu]

sys/arm/allwinner/clkng/aw_ccung.h

40 ↗

(On Diff #35675)

We should do an enum instead if we always define them.

[kevans]

sys/arm/allwinner/clkng/aw_ccung.h

40 ↗

(On Diff #35675)

*nod* Do we have a preference on explicitly defining only the first item in the enum vs. explicitly defining all of them?

[manu]

sys/arm/allwinner/clkng/aw_ccung.h

40 ↗

(On Diff #35675)

IIRC style(9) say that we only need to define the first.

[kevans]

Convert to an enum; I ultimately couldn't find anything explicit or implicit in style(9) about it, but we do only define the first in other parts of arm/allwinner so it fits the area.

[kevans]

I was going to go a step further and convert sc->type, but quickly realized that would break down and force us to move all of the individual *_ccu inclusions up into global allwinner stuff or create an even nastier typedef mess in aw_ccung.c