Re: [RFC][PATCH v4 12/69] teach handle_mounts() to handle RCU mode

[Al Viro]

On Fri, Mar 13, 2020 at 05:28:12PM -0700, Linus Torvalds wrote:
> Oh, and here you accidentally fix the problem I pointed out about
> patch 11, as you move the code:
> 
> On Fri, Mar 13, 2020 at 4:54 PM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> >
> > +               if (unlikely(!*inode))
> > +                       return -ENOENT;
> 
> Correct test added.
> 
> > -                       if (unlikely(!inode))
> > -                               return -ENOENT;
> 
> Incorrect test removed.
> 
> And again, maybe I'm misreading the patch. But it does look like it's
> wrong in the middle of the series, which would make bisection if
> there's some related bug "interesting".

Bisect hazard on botched reordering, actually.  Fixed (IOW, that should've
been if (!*inode) already in #11).