Re: [lkp] [nfsd] b44061d0b9: BUG: Dentry ffff880027d7c540{i=1846f,n=0a} still in use (1) [unmount of btrfs vda]

From: "J. Bruce Fields" <bfields@redhat.com>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Josef Bacik <jbacik@fb.com>,
	kernel test robot <xiaolong.ye@intel.com>,
	Al Viro <viro@zeniv.linux.org.uk>, Chris Mason <clm@fb.com>,
	David Sterba <dsterba@suse.com>,
	LKML <linux-kernel@vger.kernel.org>, LKP <lkp@01.org>,
	linux-btrfs <linux-btrfs@vger.kernel.org>,
	Linux NFS Mailing List <linux-nfs@vger.kernel.org>
Subject: Re: [lkp] [nfsd] b44061d0b9: BUG: Dentry ffff880027d7c540{i=1846f,n=0a} still in use (1) [unmount of btrfs vda]
Date: Wed, 10 Aug 2016 15:09:16 -0400	[thread overview]
Message-ID: <20160810190915.GA22411@pad> (raw)
In-Reply-To: <CA+55aFw5DczMkEYP47h=-9f6U2V5aQUhSxqPjtTbNp1Dd=gZwQ@mail.gmail.com>

On Wed, Aug 10, 2016 at 11:56:15AM -0700, Linus Torvalds wrote:
> On Wed, Aug 10, 2016 at 11:46 AM, Josef Bacik <jbacik@fb.com> wrote:
> >
> > So my naive fix would be something like this
> 
> Bruce? Josef's patch looks ObviouslyCorrect(tm) to me now that I look
> at it - all the other callers of fh_compose() also seem to just drop
> the dentry unconditionally, knowing that fh_compose() took a ref to it
> if needed.
> 
> In fact, the only thing I'd do differently would be to not even put
> the comment there at all, since this call site isn't any different
> from any of the others. If anything, it could go on fh_compose() if we
> want to add comments.

Yep, makes sense to me too.

OK with me if you want to take it, otherwise I'll run it through my
usual tests and send you a pull request probably today or tomorrow.

Thanks, Josef!  (And kernel test robot folks--the speed with which
they're catching stuff, and bisecting down to individual commits, is
really amazing to me.)

--b.