On Wed, Sep 04, 2024 at 06:37:01PM +0900, Kyotaro Horiguchi wrote:
> I doubt that overprotection is always better, but in this case, it's
> not overprotection at all. The flush callbacks are called
> unconditionally once we decide to flush anything. Sorry for the noise.
Yes, it's intended to be a fastpath, with checks happening if we have
no data pending for any variable-numbered stats.
--
Michael