Re: [PATCH v2] fs: Do not check if there is a fsnotify watcher on pseudo inodes

From: Amir Goldstein <amir73il@gmail.com>
To: Mel Gorman <mgorman@techsingularity.net>
Cc: Jan Kara <jack@suse.cz>, Alexander Viro <viro@zeniv.linux.org.uk>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>,
	linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v2] fs: Do not check if there is a fsnotify watcher on pseudo inodes
Date: Mon, 15 Jun 2020 21:32:16 +0300	[thread overview]
Message-ID: <CAOQ4uxikAD0FFZdnkd_aHfst0G3j0Gt1_oGDb75z8gHpaE3ERg@mail.gmail.com> (raw)
In-Reply-To: <20200615172545.GG3183@techsingularity.net>

On Mon, Jun 15, 2020 at 8:25 PM Mel Gorman <mgorman@techsingularity.net> wrote:
>
> On Mon, Jun 15, 2020 at 07:26:38PM +0300, Amir Goldstein wrote:
> > On Mon, Jun 15, 2020 at 3:14 PM Mel Gorman <mgorman@techsingularity.net> wrote:
> > >
> > > Changelog since v1
> > > o Updated changelog
> >
> > Slipped to commit message
> >
>
> It's habit, it's the layout I generally use for mm even though others
> prefer having it below ---. I wasn't sure of fsnotify's preferred format
> for tracking major differences between versions.
>
> > >
> > > The kernel uses internal mounts created by kern_mount() and populated
> > > with files with no lookup path by alloc_file_pseudo for a variety of
> > > reasons. An example of such a mount is for anonymous pipes. For pipes,
> > > every vfs_write regardless of filesystem, fsnotify_modify() is called to
> > > notify of any changes which incurs a small amount of overhead in fsnotify
> > > even when there are no watchers. It can also trigger for reads and readv
> > > and writev, it was simply vfs_write() that was noticed first.
> > >
> > > A patch is pending that reduces, but does not eliminte, the overhead of
> >
> > typo: eliminte
> >
>
> Yes.
>
> > > fsnotify but for files that cannot be looked up via a path, even that
> > > small overhead is unnecessary. The user API for fanotify is based on
> > > the pathname and a dirfd and proc entries appear to be the only visible
> > > representation of the files. Proc does not have the same pathname as the
> > > internal entry and the proc inode is not the same as the internal inode
> > > so even if fanotify is used on a file under /proc/XX/fd, no useful events
> > > are notified.
> > >
> >
> > Note that fanotify is not the only uapi to add marks, but this is fine by me
> > I suppose if Jan wants to he can make small corrections on commit.
> >
>
> True but I didn't think inotify was materially different as it also takes
> a path. Is that wrong or are there others that matter and can attach to
> a file that cannot be looked up via a path?

There are kernel/audit* and nfsd/filecache.c users, but as far as I could
tell, there is no danger from there. I was just pointing out that the fanotify
uapi argument alone is not a full proof.

Thanks,
Amir.