From mboxrd@z Thu Jan 1 00:00:00 1970 From: Glauber Costa Subject: Re: [PATCH v10 08/35] list: add a new LRU list type Date: Thu, 6 Jun 2013 15:47:29 +0400 Message-ID: <51B076D1.20007@parallels.com> References: <1370287804-3481-1-git-send-email-glommer@openvz.org> <1370287804-3481-9-git-send-email-glommer@openvz.org> <20130605160758.19e854a6995e3c2a1f5260bf@linux-foundation.org> <20130606024909.GP29338@dastard> <20130605200554.d4dae16f.akpm@linux-foundation.org> <20130606044426.GX29338@dastard> <20130606000409.e4333f7c.akpm@linux-foundation.org> <51B05069.5070404@parallels.com> <20130606025517.8400c279.akpm@linux-foundation.org> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit Cc: Dave Chinner , Glauber Costa , , Mel Gorman , , , , Michal Hocko , Johannes Weiner , , Greg Thelen , Dave Chinner To: Andrew Morton Return-path: In-Reply-To: <20130606025517.8400c279.akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org> Sender: cgroups-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-fsdevel.vger.kernel.org On 06/06/2013 01:55 PM, Andrew Morton wrote: > On Thu, 6 Jun 2013 13:03:37 +0400 Glauber Costa wrote: > >> I do think it would be more appropriate to >>> discard the lib/ idea and move it all into fs/ or mm/. >> I have no particular love for this in lib/ >> >> Most of the users are in fs/, so I see no point in mm/ >> So for me, if you are really not happy about lib, I would suggest moving >> this to fs/ > > Always feel free to differ but yes, fs/ seems better to me. > > I suggested mm/ also because that's where the shrinker core resides. > As I said, unless Dave has a strong point against it, I don't really care if it lives in lib/ or not. It is infrastructure, but not necessarily lib-like infrastructure. Now, I have been thinking about this during the last hour, and as much as all users are in fs/, putting it into mm/ would give us quite some other advantage: namely, it has been already detected that we would like to have, if possible, stronger ties between shrinkers, caches and the underlying lists. We use a bunch of mm/ infrastructure, etc. This is always something we can change if it really hurts, but right now I am 51 % mm/ 49 % fs/