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, 3 Aug 2022 09:28:19 -0400
From:   Johannes Weiner <hannes@cmpxchg.org>
To:     Andrew Morton <akpm@linux-foundation.org>
Cc:     Mel Gorman <mgorman@techsingularity.net>,
        Hugh Dickins <hughd@google.com>,
        Joonsoo Kim <iamjoonsoo.kim@lge.com>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, kernel-team@fb.com
Subject: Re: [PATCH] mm: vmscan: fix extreme overreclaim and swap floods
Message-ID: <Yup386Bq8Ek1VHJq@cmpxchg.org>
References: <20220802162811.39216-1-hannes@cmpxchg.org>
 <20220802170619.250e1a0b475222a82e7077b3@linux-foundation.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220802170619.250e1a0b475222a82e7077b3@linux-foundation.org>
Precedence: bulk

On Tue, Aug 02, 2022 at 05:06:19PM -0700, Andrew Morton wrote:
> On Tue,  2 Aug 2022 12:28:11 -0400 Johannes Weiner <hannes@cmpxchg.org> wrote:
> 
> > Change the code such that only one SWAP_CLUSTER_MAX-sized nudge toward
> > the larger LRU lists is made before bailing out on a met reclaim goal.
> 
> It seems rash to jam this into 5.20-rc1 at this stage.  I'm thinking
> 5.21-rc1 with a cc:stable?

Yeah, 5.20-rc1 sounds fast. Let's wait for reviews first and see how
much confidence we get on that change. I can't help but feel, reading
logs and comments (commit 1a501907bbea8e6ebb0b16cf6db9e9cbf1d2c813),
that my fix is how the code was intended to work from the start.

5.21 does sound a biiiit on the long side for fixing such extreme
misbehavior, IMO, once it's proven to affect production workloads in
the wild. I was hoping -rc2 or so...