From mboxrd@z Thu Jan 1 00:00:00 1970 From: Waiman Long Subject: Re: [PATCH] mm, slab: Extend slab/shrink to shrink all the memcg caches Date: Wed, 3 Jul 2019 11:14:28 -0400 Message-ID: References: <20190702183730.14461-1-longman@redhat.com> <20190703065628.GK978@dhcp22.suse.cz> <9ade5859-b937-c1ac-9881-2289d734441d@redhat.com> <20190703143701.GR978@dhcp22.suse.cz> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20190703143701.GR978@dhcp22.suse.cz> Content-Language: en-US Sender: linux-kernel-owner@vger.kernel.org List-ID: Content-Type: text/plain; charset="us-ascii" To: Michal Hocko Cc: Christoph Lameter , Pekka Enberg , David Rientjes , Joonsoo Kim , Andrew Morton , Alexander Viro , Jonathan Corbet , Luis Chamberlain , Kees Cook , Johannes Weiner , Vladimir Davydov , linux-mm@kvack.org, linux-doc@vger.kernel.org, linux-fsdevel@vger.kernel.org, cgroups@vger.kernel.org, linux-kernel@vger.kernel.org, Roman Gushchin , Shakeel Butt , Andrea Arcangeli On 7/3/19 10:37 AM, Michal Hocko wrote: > On Wed 03-07-19 09:12:13, Waiman Long wrote: >> On 7/3/19 2:56 AM, Michal Hocko wrote: >>> On Tue 02-07-19 14:37:30, Waiman Long wrote: >>>> Currently, a value of '1" is written to /sys/kernel/slab//shrink >>>> file to shrink the slab by flushing all the per-cpu slabs and free >>>> slabs in partial lists. This applies only to the root caches, though. >>>> >>>> Extends this capability by shrinking all the child memcg caches and >>>> the root cache when a value of '2' is written to the shrink sysfs file. >>> Why do we need a new value for this functionality? I would tend to think >>> that skipping memcg caches is a bug/incomplete implementation. Or is it >>> a deliberate decision to cover root caches only? >> It is just that I don't want to change the existing behavior of the >> current code. It will definitely take longer to shrink both the root >> cache and the memcg caches. > Does that matter? To whom and why? I do not expect this interface to be > used heavily. The only concern that I can see is the fact that I need to take the slab_mutex when iterating the memcg list to prevent concurrent modification. That may have some impact on other applications running in the system. However, I can put a precaution statement on the user-doc to discuss the potential performance impact. >> If we all agree that the only sensible >> operation is to shrink root cache and the memcg caches together. I am >> fine just adding memcg shrink without changing the sysfs interface >> definition and be done with it. > The existing documentation is really modest on the actual semantic: > Description: > The shrink file is written when memory should be reclaimed from > a cache. Empty partial slabs are freed and the partial list is > sorted so the slabs with the fewest available objects are used > first. > > which to me sounds like all slabs are free and nobody should be really > thinking of memcgs. This is simply drop_caches kinda thing. We surely do > not want to drop caches only for the root memcg for /proc/sys/vm/drop_caches > right? > I am planning to reword the document to make the effect of using this sysfs file more explicit. Cheers, Longman