At Sun, 19 Jun 2022 22:55:00 +0900, Izumi Tsutsui wrote: > I guess essencial part is "shift op" vs "log2 calculation" for blocksize. > The logic seems ok but there are several comments on implementation: > > - maybe it's better to have function prototypes, even if they are static > - I wonder if human2blk(), human2bsd(), and bsd2blk() should be unsigned Right. I will do it. > - is it worth to explicitly specify static inline (or gcc already does?) gcc already does. Furthermore, more large check_unit() or readdisklabel() are also inlined. Therefore I won't add inline keyword for these new functions. > - if now cdstrategy() is identical with sdstrategy(), maybe it's better > to rename current sdstrategy() to sdcdstrategy() and use it both > sd and cd in devsw[] in stand/boot/conf.c? I've tried a bit but there are several things to consider about it. I won't do it at this time. Thank you for reviewing. I will commit it. --- Tetsuya Isaki <isaki%pastel-flower.jp@localhost / isaki%NetBSD.org@localhost>