Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:4 as permitted sender) client-ip=2620:137:e000::3:4;
MIME-Version: 1.0
References: <20231006160024.170748-1-hannes@cmpxchg.org>
In-Reply-To: <20231006160024.170748-1-hannes@cmpxchg.org>
From:   Nhat Pham <nphamcs@gmail.com>
Date:   Fri, 6 Oct 2023 14:40:48 -0700
Message-ID: <CAKEwX=N_8ZORaiUDrGQON7RFgC0mXCyyHhR9s9Zki1qONEHnKw@mail.gmail.com>
Subject: Re: [PATCH] mm: zswap: fix pool refcount bug around shrink_worker()
To:     Johannes Weiner <hannes@cmpxchg.org>
Cc:     Andrew Morton <akpm@linux-foundation.org>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org, Chris Mason <clm@fb.com>,
        stable@vger.kernel.org, Vitaly Wool <vitaly.wool@konsulko.com>,
        Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Fri, Oct 6, 2023 at 9:00=E2=80=AFAM Johannes Weiner <hannes@cmpxchg.org>=
 wrote:
>
> When a zswap store fails due to the limit, it acquires a pool
> reference and queues the shrinker. When the shrinker runs, it drops
> the reference. However, there can be multiple store attempts before
> the shrinker wakes up and runs once. This results in reference leaks
> and eventual saturation warnings for the pool refcount.
>
> Fix this by dropping the reference again when the shrinker is already
> queued. This ensures one reference per shrinker run.
>
> Reported-by: Chris Mason <clm@fb.com>
> Fixes: 45190f01dd40 ("mm/zswap.c: add allocation hysteresis if pool limit=
 is hit")
> Cc: stable@vger.kernel.org      [5.6+]
> Cc: Vitaly Wool <vitaly.wool@konsulko.com>
> Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
> Cc: Nhat Pham <nphamcs@gmail.com>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
> ---
>  mm/zswap.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/zswap.c b/mm/zswap.c
> index 083c693602b8..37d2b1cb2ecb 100644
> --- a/mm/zswap.c
> +++ b/mm/zswap.c
> @@ -1383,8 +1383,8 @@ bool zswap_store(struct folio *folio)
>
>  shrink:
>         pool =3D zswap_pool_last_get();
> -       if (pool)
> -               queue_work(shrink_wq, &pool->shrink_work);
> +       if (pool && !queue_work(shrink_wq, &pool->shrink_work))
> +               zswap_pool_put(pool);
>         goto reject;
>  }
>
> --
> 2.42.0
>

Acked-by: Nhat Pham <nphamcs@gmail.com>

Random tangent: this asynchronous writeback mechanism
is always kinda weird to me. We could have quite a bit of memory
inversion before the shrinker finally kicks in and frees up zswap
pool space. But I guess if it doesn't break then don't fix it.

Maybe a shrinker that proactively writes pages back as memory
pressure builds up could help ;)