On Mon, Mar 30, 2020 at 03:33:31PM +0200, Jan Kara wrote: > Honestly, I think that a microoptimization like this will bring any benefit > only in the fast paths and e.g. most of the filesystem places aren't like > that as far as I've checked. OTOH it doesn't hurt readability so I'm not > opposed either. So feel free to add: > > Acked-by: Jan Kara <jack@xxxxxxx> > > for ext2 and ext4 bits. Thank you! I fully agree with you, but - Figuring out whether a particular site is a fast path or not is a lot more work than simply fixing it, - It's also smaller, which is what we care about for slow-path code. and - I wish people would stop copying the slow code to fast paths. If I make sure the kernel is full of good examples and devoid of bad examples, the odds improve.
