Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 23 Feb 2022 16:10:44 +0900
From:   Jonghyeon Kim <tome01@ajou.ac.kr>
To:     SeongJae Park <sj@kernel.org>
Cc:     Jonghyeon Kim <tome01@ajou.ac.kr>, Jonathan.Cameron@Huawei.com,
        amit@kernel.org, benh@kernel.crashing.org, corbet@lwn.net,
        david@redhat.com, dwmw@amazon.com, elver@google.com,
        foersleo@amazon.de, gthelen@google.com, markubo@amazon.de,
        rientjes@google.com, shakeelb@google.com, shuah@kernel.org,
        linux-damon@amazon.com, linux-mm@kvack.org,
        linux-doc@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [RFC PATCH v1 1/3] mm/damon: Rebase damos watermarks for NUMA
 systems
Message-ID: <20220223071044.GA6270@swarm08>
References: <20220218102611.31895-2-tome01@ajou.ac.kr>
 <20220222095317.7911-1-sj@kernel.org>
 <20220223051056.GA3502@swarm08>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220223051056.GA3502@swarm08>
User-Agent: Mutt/1.9.4 (2018-02-28)
Precedence: bulk

On Wed, Feb 23, 2022 at 02:10:56PM +0900, Jonghyeon Kim wrote:
> On Tue, Feb 22, 2022 at 09:53:17AM +0000, SeongJae Park wrote:
> > Hello Jonghyeon,
> > 
> 
> Hello SeongJae,
> 
> > On Fri, 18 Feb 2022 19:26:09 +0900 Jonghyeon Kim <tome01@ajou.ac.kr> wrote:
> > 
> > > For NUMA systems, there is a need to allow damos to select watermark
> > > options for monitoring each NUMA node or whole system free memory. Even
> > > if we do not use NUMA, since the default NUMA node number is 0, we can
> > > monitor the whole system memory without any configuration.
> > 
> > Some users using NUMA machines but don't do NUMA-specific memory allocations
> > and therefore assume memory free rate in each NUMA node will be similar might
> > want to monitor only global free memory ratio, to limit number of kdamonds for
> > reducing CPU overhead.  In the case, this patch would make them monitor only
> > the first node.
> 
> I think that the purpose of DAMON_RECALIM is proactively reclaiming memory
> before waking up kswapd. But I missed that case you mentioned. In that case,
> they might not want to monitor each NUMA node. I agree with this.
> 
> > 
> > How about leaving DAMOS_WMARK_FREE_MEM_RATE to work as is, and adding a new
> > metric type, say, DAMOS_WMARK_NODE_FREE_MEM_RATE?
> > 
> 
> Yes, it looks good to me.
> 
> 
> Thanks,
> Jonghyeon