Re: [PATCH 2/2] fsnotify: turn fsnotify reaper thread into a workqueue job

From: Jeff Layton <jlayton@poochiereds.net>
To: Eryu Guan <guaneryu@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>,
	Jan Kara <jack@suse.com>, Eric Paris <eparis@parisplace.org>
Subject: Re: [PATCH 2/2] fsnotify: turn fsnotify reaper thread into a workqueue job
Date: Mon, 15 Feb 2016 12:55:22 -0500	[thread overview]
Message-ID: <20160215125522.03865cf4@tlielax.poochiereds.net> (raw)
In-Reply-To: <20160215060310.GD29898@eguan.usersys.redhat.com>

On Mon, 15 Feb 2016 14:03:10 +0800
Eryu Guan <guaneryu@gmail.com> wrote:

> On Sun, Feb 14, 2016 at 07:15:23PM -0500, Jeff Layton wrote:
> > We don't require a dedicated thread for fsnotify cleanup. Switch it over
> > to a workqueue job instead that runs on the system_unbound_wq.
> > 
> > In the interest of not thrashing the queued job too often when there are
> > a lot of marks being removed, we delay the reaper job slightly when
> > queueing it, to allow several to gather on the list.
> > 
> > Cc: Jan Kara <jack@suse.com>
> > Cc: Eric Paris <eparis@parisplace.org>
> > Cc: Andrew Morton <akpm@linux-foundation.org>
> > Cc: Eryu Guan <guaneryu@gmail.com>
> > Signed-off-by: Jeff Layton <jeff.layton@primarydata.com>  
> 
> With these two patches applied on top of r.5-rc3, the same test passed
> 2-hour stress run, also survived stress test that forks 5 processes
> running the same test program for 30 minutes.
> 
> Thanks for looking into this!
> 
> Eryu

Thanks for reporting and testing the patches. Yes, I tested them too
while running your reproducer and watched the size of the
inotify_inode_mark slab. It seemed to stay pretty stable with these
patches as well.

Andrew, would you mind picking up these patches since you merged the
original one?

Thanks,
-- 
Jeff Layton <jlayton@poochiereds.net>