From: Vladimir Davydov Subject: Re: [Devel] [PATCH v11 00/15] kmemcg shrinkers Date: Tue, 26 Nov 2013 16:55:43 +0400 Message-ID: <52949A4F.7030004@parallels.com> References: <20131125174135.GE22729@cmpxchg.org> <529443E4.7080602@parallels.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <529443E4.7080602-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org> Sender: cgroups-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: text/plain; charset="us-ascii"; format="flowed" To: Johannes Weiner Cc: linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, glommer-GEFAQzZX7r8dnm+yROfE0A@public.gmane.org, mhocko-AlSwsSmVLrQ@public.gmane.org, linux-mm-Bw31MaZKKs3YtjvyW6yDsg@public.gmane.org, cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org, devel-GEFAQzZX7r8dnm+yROfE0A@public.gmane.org On 11/26/2013 10:47 AM, Vladimir Davydov wrote: > Hi, > > Thank you for the review. I agree with all your comments and I'll > resend the fixed version soon. > > If anyone still has something to say about the patchset, I'd be glad > to hear from them. > > On 11/25/2013 09:41 PM, Johannes Weiner wrote: >> I ran out of steam reviewing these because there were too many things >> that should be changed in the first couple patches. >> >> I realize this is frustrating to see these type of complaints in v11 >> of a patch series, but the review bandwidth was simply exceeded back >> when Glauber submitted this along with the kmem accounting patches. A >> lot of the kmemcg commits themselves don't even have review tags or >> acks, but it all got merged anyway, and the author has moved on to >> different projects... >> >> Too much stuff slips past the only two people that have more than one >> usecase on their agenda and are willing to maintain this code base - >> which is in desparate need of rework and pushback against even more >> drive-by feature dumps. I have repeatedly asked to split the memcg >> tree out of the memory tree to better deal with the vastly different >> developmental stages of memcg and the rest of the mm code, to no >> avail. So I don't know what to do anymore, but this is not working. >> >> Thoughts? > > That's a pity, because w/o this patchset kmemcg is in fact useless. > Perhaps, it's worth trying to split it? (not sure if it'll help much > though since first 11 patches are rather essential :-( ) What do you think about splitting this set into two main series as follows: 1) Prepare vmscan to kmemcg-aware shrinkers; would include patches 1-7 of this set. 2) Make fs shrinkers memcg-aware; would include patches 9-11 of this set and leave other patches, which are rather for optimization/extending functionality, for future? Thanks.