Re: [PATCH] vfs: fsmount: add missing mntget()

[Eric Biggers <ebiggers@kernel.org>]

On Wed, Jul 10, 2019 at 01:31:13AM +0100, Al Viro wrote:
> On Tue, Jul 09, 2019 at 04:00:29PM -0700, Eric Biggers wrote:
> 
> > > index 49a058c73e4c..26f74e092bd9 100644
> > > --- a/fs/pnode.h
> > > +++ b/fs/pnode.h
> > > @@ -44,7 +44,7 @@ int propagate_mount_busy(struct mount *, int);
> > >  void propagate_mount_unlock(struct mount *);
> > >  void mnt_release_group_id(struct mount *);
> > >  int get_dominating_id(struct mount *mnt, const struct path *root);
> > > -unsigned int mnt_get_count(struct mount *mnt);
> > > +int mnt_get_count(struct mount *mnt);
> > >  void mnt_set_mountpoint(struct mount *, struct mountpoint *,
> > >  			struct mount *);
> > >  void mnt_change_mountpoint(struct mount *parent, struct mountpoint *mp,
> > 
> > Miklos, are you planning to send this as a formal patch?
> 
> Hold it for a while, OK?  There's an unpleasant issue (a very long-standing
> one) with boxen that have an obscene amount of RAM.  Some of the counters
> involved will need to become long.  This is the coming cycle fodder (mounts
> and inodes are relatively easy; it's dentry->d_count that brings arseloads
> of fun) and I'd rather deal with that sanity check as part of the same series.
> It's not forgotten...  Patch series re limiting the number of negative
> dentries is also getting into the same mix.  Watch #work.dcache - what's
> in there is basically prep work for the big pile for the next cycle; it'll
> be interesting...

Al, whatever happened to the refcounting patches you mentioned here?

- Eric