Received-SPF: pass (google.com: domain of linux-kernel+bounces-150730-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Precedence: bulk
MIME-Version: 1.0
References: <3iccc6vjl5gminut3lvpl4va2lbnsgku5ei2d7ylftoofy3n2v@gcfdvtsq6dx2>
 <CAKEwX=MZ3jTVpN4g-qrhTn2b0i0C6_M=8BtKt9KEPyFHb+4W2w@mail.gmail.com>
 <CAKEwX=NM1y-K1-Yw=CH3cM-8odER1PZBVoWo-rs7_OdjFG_puw@mail.gmail.com>
 <CAKEwX=MWPUf1NMGdn+1AkRdOUf25ifAbPyoP9zppPTx3U3Tv2Q@mail.gmail.com>
 <246c1f4d-af13-40fa-b968-fbaf36b8f91f@linux.dev> <20240417143324.GA1055428@cmpxchg.org>
 <4c3ppfjxnrqx6g52qvvhqzcc4pated2q5g4mi32l22nwtrkqfq@a4lk6s5zcwvb> <20240418124043.GC1055428@cmpxchg.org>
In-Reply-To: <20240418124043.GC1055428@cmpxchg.org>
From: Yosry Ahmed <yosryahmed@google.com>
Date: Thu, 18 Apr 2024 13:09:22 -0700
Message-ID: <CAJD7tkaPMQqQtfxcLWraz-vnbAxZKxuJRJ7vKuDOCCXtpBSF1A@mail.gmail.com>
Subject: Re: [REGRESSION] Null pointer dereference while shrinking zswap
To: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christian Heusel <christian@heusel.eu>, Chengming Zhou <chengming.zhou@linux.dev>, 
	Nhat Pham <nphamcs@gmail.com>, Seth Jennings <sjenning@redhat.com>, 
	Dan Streetman <ddstreet@ieee.org>, Vitaly Wool <vitaly.wool@konsulko.com>, 
	Andrew Morton <akpm@linux-foundation.org>, linux-mm@kvack.org, 
	linux-kernel@vger.kernel.org, David Runge <dave@sleepmap.de>, 
	"Richard W.M. Jones" <rjones@redhat.com>, Mark W <instruform@gmail.com>, regressions@lists.linux.dev
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Apr 18, 2024 at 5:40=E2=80=AFAM Johannes Weiner <hannes@cmpxchg.org=
> wrote:
>
> On Wed, Apr 17, 2024 at 07:18:14PM +0200, Christian Heusel wrote:
> > On 24/04/17 10:33AM, Johannes Weiner wrote:
> > >
> > > Christian, can you please test the below patch on top of current
> > > upstream?
> > >
> >
> > Hey Johannes,
> >
> > I have applied your patch on top of 6.9-rc4 and it did solve the crash =
for
> > me, thanks for hacking together a fix so quickly! =F0=9F=A4=97
> >
> > Tested-By: Christian Heusel <christian@heusel.eu>
>
> Thanks for confirming it, and sorry about the breakage.
>
> Andrew, can you please use the updated changelog below?
>
> ---
>
> From 52f67f5fab6a743c2aedfc8e04a582a9d1025c28 Mon Sep 17 00:00:00 2001
> From: Johannes Weiner <hannes@cmpxchg.org>
> Date: Thu, 18 Apr 2024 08:26:28 -0400
> Subject: [PATCH] mm: zswap: fix shrinker NULL crash with cgroup_disable=
=3Dmemory
>
> Christian reports a NULL deref in zswap that he bisected down to the
> zswap shrinker. The issue also cropped up in the bug trackers of
> libguestfs [1] and the Red Hat bugzilla [2].
>
> The problem is that when memcg is disabled with the boot time flag,
> the zswap shrinker might get called with sc->memcg =3D=3D NULL. This is
> okay in many places, like the lruvec operations. But it crashes in
> memcg_page_state() - which is only used due to the non-node accounting
> of cgroup's the zswap memory to begin with.
>
> Nhat spotted that the memcg can be NULL in the memcg-disabled case,
> and I was then able to reproduce the crash locally as well.
>
> [1] https://github.com/libguestfs/libguestfs/issues/139
> [2] https://bugzilla.redhat.com/show_bug.cgi?id=3D2275252
>
> Fixes: b5ba474f3f51 ("zswap: shrink zswap pool based on memory pressure")
> Cc: stable@vger.kernel.org      [v6.8]
> Link: https://lkml.kernel.org/r/20240417143324.GA1055428@cmpxchg.org
> Reported-by: Christian Heusel <christian@heusel.eu>
> Debugged-by: Nhat Pham <nphamcs@gmail.com>
> Suggested-by: Nhat Pham <nphamcs@gmail.com>
> Tested-By: Christian Heusel <christian@heusel.eu>
> Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>

Thanks for fixing this. A couple of comments/questions below, but anyway LG=
TM:

Acked-by: Yosry Ahmed <yosryahmed@google.com>

> ---
>  mm/zswap.c | 25 ++++++++++++++++---------
>  1 file changed, 16 insertions(+), 9 deletions(-)
>
> diff --git a/mm/zswap.c b/mm/zswap.c
> index caed028945b0..6f8850c44b61 100644
> --- a/mm/zswap.c
> +++ b/mm/zswap.c
> @@ -1331,15 +1331,22 @@ static unsigned long zswap_shrinker_count(struct =
shrinker *shrinker,
>         if (!gfp_has_io_fs(sc->gfp_mask))
>                 return 0;
>
> -#ifdef CONFIG_MEMCG_KMEM
> -       mem_cgroup_flush_stats(memcg);
> -       nr_backing =3D memcg_page_state(memcg, MEMCG_ZSWAP_B) >> PAGE_SHI=
FT;
> -       nr_stored =3D memcg_page_state(memcg, MEMCG_ZSWAPPED);
> -#else
> -       /* use pool stats instead of memcg stats */
> -       nr_backing =3D zswap_pool_total_size >> PAGE_SHIFT;
> -       nr_stored =3D atomic_read(&zswap_nr_stored);
> -#endif
> +       /*
> +        * For memcg, use the cgroup-wide ZSWAP stats since we don't
> +        * have them per-node and thus per-lruvec. Careful if memcg is
> +        * runtime-disabled: we can get sc->memcg =3D=3D NULL, which is o=
k
> +        * for the lruvec, but not for memcg_page_state().
> +        *
> +        * Without memcg, use the zswap pool-wide metrics.
> +        */
> +       if (!mem_cgroup_disabled()) {

With the current shrinker code, it seems like we cannot get sc->memcg
=3D=3D NULL unless mem_cgroup_disabled() is true indeed. However, maybe
it's better to check if sc->memcg is not NULL directly instead?

This would be more resilient in case the shrinker code changes and
passing sc->memcg =3D=3D NULL becomes possible in other cases (e.g. during
global shrinking). It seems like other shrinkers do this, for example
see count_shadow_nodes() and deferred_split_count().

I am also wondering if we should also check !mem_cgroup_is_root()
here? We can avoid the expensive global flush and use the global stats
directly in this case. I could also send a follow up patch for this if
that's preferred.

> +               mem_cgroup_flush_stats(memcg);
> +               nr_backing =3D memcg_page_state(memcg, MEMCG_ZSWAP_B) >> =
PAGE_SHIFT;
> +               nr_stored =3D memcg_page_state(memcg, MEMCG_ZSWAPPED);
> +       } else {
> +               nr_backing =3D zswap_pool_total_size >> PAGE_SHIFT;
> +               nr_stored =3D atomic_read(&zswap_nr_stored);
> +       }
>
>         if (!nr_stored)
>                 return 0;
> --
> 2.44.0