David Howells wrote on Mon, Sep 20, 2021 at 01:14:15PM +0100: > Deal with some warnings generated from make W=1: > > (1) Add/remove/fix kerneldoc parameters descriptions. > > (2) afs_sillyrename() isn't an API functions, so remove the kerneldoc > annotation. > > (3) The fscache object CREATE_OBJECT work state isn't used, so remove it. > > (4) Move __add_fid() from between v9fs_fid_add() and its comment. > > (5) 9p's caches_show() doesn't really make sense as an API function, show > remove the kerneldoc annotation. It's also not prefixed with 'v9fs_'. Happy with the 9p changes: Reviewed-by: Dominique Martinet <asmadeus@xxxxxxxxxxxxx> Having all of these in a single commit makes it difficult to deal but I don't expect any conflict on my end, so happy to have it go in your fscache tree. Matthew Wilcox wrote on Mon, Sep 20, 2021 at 01:37:46PM +0100: > This is an example of a weird pattern in filesystems. Several of > them have kernel-doc for the implementation of various ->ops methods. > I don't necessarily believe we should delete the comments (although is > there any useful information in the above?), but I don't see the point > in the comment being kernel-doc. As far as I'm concerned this is just an "it's always been like this" thing for me/9p, I wouldn't mind if it were all converted to normal comments -- but now it's describing arguments by name having it as kerneldoc has helped catch comments which didn't get updated when function changed quite a few times in patches similar to this one so it would only make sense if we remove obvious argument descriptions as well in my opinion, and that's a bit of manual work. -- Dominique
