Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:5 as permitted sender) client-ip=2620:137:e000::3:5;
Date:   Fri, 17 Nov 2023 16:19:56 +0000
From:   Mel Gorman <mgorman@techsingularity.net>
To:     Michal Hocko <mhocko@suse.com>
Cc:     Charan Teja Kalla <quic_charante@quicinc.com>,
        akpm@linux-foundation.org, david@redhat.com, vbabka@suse.cz,
        hannes@cmpxchg.org, quic_pkondeti@quicinc.com, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH V2 2/3] mm: page_alloc: correct high atomic reserve
 calculations
Message-ID: <20231117161956.d3yjdxhhm4rhl7h2@techsingularity.net>
References: <cover.1699104759.git.quic_charante@quicinc.com>
 <905d99651423ee85aeb7a71982b95ee9bb05ee99.1699104759.git.quic_charante@quicinc.com>
 <20231116095901.72oqtjoiqzt3tfhj@techsingularity.net>
 <ZVYQihO8lsI5_JJc@tiehlicka>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-15
Content-Disposition: inline
In-Reply-To: <ZVYQihO8lsI5_JJc@tiehlicka>
Precedence: bulk

On Thu, Nov 16, 2023 at 01:52:26PM +0100, Michal Hocko wrote:
> On Thu 16-11-23 09:59:01, Mel Gorman wrote:
> [...]
> > This patch in isolation seems fine with the caveat that such a small
> > system may find the atomic reserves to be borderline useless.
> 
> Yes, exactly what I had in mind. Would it make sense to reserve the the
> pageblock only if it really is less than 1% of available memory?

I'd have no objection with the caveat that we need to watch out for new
bugs related to high-order atomic failures. If ths risk is noted in the
changelog and that it's a revert candidate then any bisection should
trivially find it.

-- 
Mel Gorman
SUSE Labs