On 26/05/2020 17:33, Nikolay Borisov wrote: [...] > Always be careful of such results since this is likely due to less > inlining. The bulk of the size overhead is likely in the new function > and now those have been replaced by a 'call' and a function prologue > etc. So this doesn't always mean better performance is all I'm trying to > say ;) Yep I know, I've just included it as a small "bonus" but my point was the increased readability. >> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@xxxxxxx> > > Reviewed-by: Nikolay Borisov <nborisov@xxxxxxxx> > > Although look down for one minor discussion point. [...] >> +static int read_bg_from_eb(struct btrfs_fs_info *fs_info, struct btrfs_key *key, >> + struct extent_buffer *leaf, int slot) > > nit: I wonder if instead of passing leaf/slot it'll be better to pass > btrfs_path, since that function always refers to nodes/slots [0]. My gut > feeling is this provides better interface abstraction, but this is > effectively a private, function so I might be overthinking it. Agreed, I'll add it in a v2.
