[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAJD7tkY3t92A2ky4788hNBBZFafNeOjy_F6+E+62mySB5UGYug@mail.gmail.com>
Date: Wed, 13 Dec 2023 20:28:53 -0800
From: Yosry Ahmed <yosryahmed@...gle.com>
To: Dan Schatzberg <schatzberg.dan@...il.com>
Cc: Yu Zhao <yuzhao@...gle.com>, Johannes Weiner <hannes@...xchg.org>,
Roman Gushchin <roman.gushchin@...ux.dev>,
Huan Yang <link@...o.com>, linux-kernel@...r.kernel.org,
cgroups@...r.kernel.org, linux-mm@...ck.org,
Tejun Heo <tj@...nel.org>, Zefan Li <lizefan.x@...edance.com>,
Jonathan Corbet <corbet@....net>,
Michal Hocko <mhocko@...nel.org>,
Shakeel Butt <shakeelb@...gle.com>,
Muchun Song <muchun.song@...ux.dev>,
Andrew Morton <akpm@...ux-foundation.org>,
David Hildenbrand <david@...hat.com>,
Matthew Wilcox <willy@...radead.org>,
Kefeng Wang <wangkefeng.wang@...wei.com>,
Yue Zhao <findns94@...il.com>, Hugh Dickins <hughd@...gle.com>
Subject: Re: [PATCH V4 2/2] mm: add swapiness= arg to memory.reclaim
On Wed, Dec 13, 2023 at 8:38 AM Dan Schatzberg <schatzberg.dan@...il.com> wrote:
>
> On Tue, Dec 12, 2023 at 07:05:36PM -0700, Yu Zhao wrote:
> > On Tue, Dec 12, 2023 at 6:39 PM Dan Schatzberg <schatzberg.dan@...il.com> wrote:
> > >
> > > Allow proactive reclaimers to submit an additional swappiness=<val>
> > > argument to memory.reclaim. This overrides the global or per-memcg
> > > swappiness setting for that reclaim attempt.
> > >
> > > For example:
> > >
> > > echo "2M swappiness=0" > /sys/fs/cgroup/memory.reclaim
> > >
> > > will perform reclaim on the rootcg with a swappiness setting of 0 (no
> > > swap) regardless of the vm.swappiness sysctl setting.
> > >
> > > Signed-off-by: Dan Schatzberg <schatzberg.dan@...il.com>
> >
> > NAK.
> >
> > Please initialize new variables properly and test code changes
> > thoroughly before submission.
>
> Could you be a bit more specific? The patch is compiling and working
> locally but perhaps there's some configuration or behavior that I
> haven't been testing.
scan_control.swappiness is only initialized from
try_to_free_mem_cgroup_pages(), which means that swappiness is now 0
for all other types of reclaim, which can be a huge problem.
It might be easier to restore a special value (-1, 201, whatever) that
means "use mem_cgroup_swappiness()".
Powered by blists - more mailing lists