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;
From:   SeongJae Park <sj@kernel.org>
To:     Pedro Gomes <pedrodemargomes@gmail.com>
Cc:     SeongJae Park <sj@kernel.org>,
        Steven Rostedt <rostedt@goodmis.org>,
        Ingo Molnar <mingo@redhat.com>,
        Andrew Morton <akpm@linux-foundation.org>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mm/damon: Add option to monitor only writes
Date:   Mon,  7 Feb 2022 08:16:25 +0000
Message-Id: <20220207081625.3579-1-sj@kernel.org>
In-Reply-To: <CACpbsfZN2fExaTReJG1anDQjCd3i24_0wG-oU42dSHD5bAon7A@mail.gmail.com>
Precedence: bulk

On Fri, 4 Feb 2022 12:11:44 -0300 Pedro Gomes <pedrodemargomes@gmail.com> wrote:

> On Thu, Feb 3, 2022 at 10:39 AM SeongJae Park <sj@kernel.org> wrote:
> 
> > I think this would be better to be implemented as another monitoring primitive
> > based on the virtual address space monitoring primitives, e.g., vaddr-writes?
> > Then the implementation would be simpler and changes to other files will be
> > minimized.  For the user space interface, we could use a prefix to target_ids
> > debugfs file.  For example,
> >
> >     # echo "vaddr-writes $(pidof workload)" > $debugfs/damon/target_ids
> 
> I will do that.

Thanks!

> 
> > > This patch also adds the actions mergeable and unmergeable to damos schemes.
> > > These actions are used by KSM as explained in [1].
> >
> > Please do that in a separate patch, and also update the document
> > (Documentation/admin-guide/mm/damon/usage.rst).  And, what's the expected usage
> > of the action and benefits?
> 
> The idea is to use this action to all areas that are not written too frequently,
> this way KSM can save more memory without too much overhead.
> But I have to test it better and collect some data to see if it really
> makes sense,
> perhaps it is better to leave this patch for later.
> I would like to know your opinion on this, do you think it makes sense?

Yes, that idea makes sense to me :)


Thanks,
SJ

> 
> 
> -- 
> Atenciosamente,
> Pedro Demarchi Gomes.