From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201])
	(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
	(No client certificate requested)
	by smtp.subspace.kernel.org (Postfix) with ESMTPS id E936D6A022
	for <linux-nfs@vger.kernel.org>; Mon,  8 Jul 2024 13:35:07 +0000 (UTC)
Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=10.30.226.201
ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116;
	t=1720445708; cv=none; b=O9Npw5Nh2jFOb1LPoWxnYo3JwVZJunUH5JleD1DqJR+yvAg0ZRVxy6H/1Ud3K6K5GG7qOnFZN/w3COTSMV51p17WZMXwUHQyQAprwSs5fYz87tpdqm2KDSdXcR4TF1FkH1JzuL5jTHY0I+9tSm1pcEvtSj3zkpe4EI1U0QzfuxE=
ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org;
	s=arc-20240116; t=1720445708; c=relaxed/simple;
	bh=QmhmXr3PA1ZV++s8Wxje+2+/LGDox6++KLGQ2DFY244=;
	h=Message-ID:Subject:From:To:Cc:Date:In-Reply-To:References:
	 Content-Type:MIME-Version; b=IwG0jsLClnaTFHlaxMQ/RyKtCip7rLc0WaCnOuf9WzB7TYFHqvUnJmYqVhDLpce8dJ6LkVfSbLeX0Ub0XszLcHtLb0LqZ4Jr885XVbng2ue0ufmdsOEZUNbXVfIDPJBIxVoWHV1h+oHJ06PRkAQ7rA4W7iUmrKl8YQgpybrSfLs=
ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=hC6i44yV; arc=none smtp.client-ip=10.30.226.201
Authentication-Results: smtp.subspace.kernel.org;
	dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="hC6i44yV"
Received: by smtp.kernel.org (Postfix) with ESMTPSA id 1AA2AC116B1;
	Mon,  8 Jul 2024 13:35:07 +0000 (UTC)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org;
	s=k20201202; t=1720445707;
	bh=QmhmXr3PA1ZV++s8Wxje+2+/LGDox6++KLGQ2DFY244=;
	h=Subject:From:To:Cc:Date:In-Reply-To:References:From;
	b=hC6i44yViiZYv7Qe9axpKCWUfOZC2sqvEFS91hSwv6Pslt0wuAKLOfgVZhn6oESM/
	 7aoaypdq7sd5yXKWjDX2vIeWJcN+x0d4R5BDbF9eQBHeLv4O8cpD8xRTj/ZxIhPuDB
	 o8o86/rm8epPNe3BaYHAjuLf1/5ccUTYbgyvnO9CtdJKR3q3W/m1ISSIck3Od84FU6
	 BOQnLyLEmDORkYFnykFggapd67vv6PUAcuwProMrcenFQFRRdeWTfDwtle5Cg++wNc
	 PrRONlrQLaWmNmLQGuWvFnREidy6tIZ/UIwt6KvHQkSjUMgGNsNwCaZgyzL6l1fW56
	 IUyHu9Z1na0GQ==
Message-ID: <321ddc16356d75f9eb6e5ab15c4e28fae1466267.camel@kernel.org>
Subject: Re: Leaked nfsd_file due to race condition and early unhash
 (fs/nfsd/filecache.c)
From: Jeff Layton <jlayton@kernel.org>
To: Youzhong Yang <youzhong@gmail.com>
Cc: Chuck Lever <chuck.lever@oracle.com>, linux-nfs@vger.kernel.org
Date: Mon, 08 Jul 2024 09:35:05 -0400
In-Reply-To: <CADpNCvZ-kEc6hOQHsbn7yHtvB-acg_gQwzEjN9zcjw0oM2RgGw@mail.gmail.com>
References: 
	<CADpNCvYGqA3a51OH=AcqmKyAmnx3yoZjYPo7US+qk-OMX789vA@mail.gmail.com>
	 <ZoWWis0AgvmiVzBU@tissot.1015granger.net>
	 <CADpNCvbxN5hmORArs+vb5D7nRC4xNf1U4oUSDbkUx8MPV547rA@mail.gmail.com>
	 <0445d64ebcc7185bf48cc05f72ca29b859f45c26.camel@poochiereds.net>
	 <CADpNCvZ-kEc6hOQHsbn7yHtvB-acg_gQwzEjN9zcjw0oM2RgGw@mail.gmail.com>
Autocrypt: addr=jlayton@kernel.org; prefer-encrypt=mutual;
 keydata=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
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.50.4 (3.50.4-1.fc39) 
Precedence: bulk
X-Mailing-List: linux-nfs@vger.kernel.org
List-Id: <linux-nfs.vger.kernel.org>
List-Subscribe: <mailto:linux-nfs+subscribe@vger.kernel.org>
List-Unsubscribe: <mailto:linux-nfs+unsubscribe@vger.kernel.org>
MIME-Version: 1.0

On Mon, 2024-07-08 at 08:58 -0400, Youzhong Yang wrote:
> Thank you Jeff for your invaluable insights. I was leaning towards
> adding a new list_head too, and tested this approach on kernel 6.6 by
> continuously hammering the server with heavy nfs load for the last few
> days, not a single leak.
>=20
> Here goes the patch (based on Linux kernel master branch), please review:
>=20
> From: Youzhong Yang <youzhong@gmail.com>
> Date: Thu, 4 Jul 2024 11:25:40 -0400
> Subject: [PATCH] nfsd: fix nfsd_file leaking due to mixed use of nf->nf_l=
ru
>=20
> nfsd_file_put() in one thread can race with another thread doing
> garbage collection (running nfsd_file_gc() -> list_lru_walk() ->
> nfsd_file_lru_cb()):
>=20
> =C2=A0 * In nfsd_file_put(), nf->nf_ref is 1, so it tries to do nfsd_file=
_lru_add().
> =C2=A0 * nfsd_file_lru_add() returns true (with NFSD_FILE_REFERENCED bit =
set)
> =C2=A0 * garbage collector kicks in, nfsd_file_lru_cb() clears REFERENCED=
 bit and
> =C2=A0=C2=A0=C2=A0 returns LRU_ROTATE.
> =C2=A0 * garbage collector kicks in again, nfsd_file_lru_cb() now
> decrements nf->nf_ref
> =C2=A0=C2=A0=C2=A0 to 0, runs nfsd_file_unhash(), removes it from the LRU=
 and adds to
> the dispose
> =C2=A0=C2=A0=C2=A0 list [list_lru_isolate_move(lru, &nf->nf_lru, head)]
> =C2=A0 * nfsd_file_put() detects NFSD_FILE_HASHED bit is cleared, so it
> tries to remove
> =C2=A0=C2=A0=C2=A0 the 'nf' from the LRU [if (!nfsd_file_lru_remove(nf))]=
. The 'nf'
> has been added
> =C2=A0=C2=A0=C2=A0 to the 'dispose' list by nfsd_file_lru_cb(), so
> nfsd_file_lru_remove(nf) simply
> =C2=A0=C2=A0=C2=A0 treats it as part of the LRU and removes it, which lea=
ds to its removal from
> =C2=A0=C2=A0=C2=A0 the 'dispose' list.
> =C2=A0 * At this moment, 'nf' is unhashed with its nf_ref being 0, and no=
t
> on the LRU.
> =C2=A0=C2=A0=C2=A0 nfsd_file_put() continues its execution [if
> (refcount_dec_and_test(&nf->nf_ref))],
> =C2=A0=C2=A0=C2=A0 as nf->nf_ref is already 0, nf->nf_ref is set to
> REFCOUNT_SATURATED, and the 'nf'
> =C2=A0=C2=A0=C2=A0 gets no chance of being freed.
>=20
> nfsd_file_put() can also race with nfsd_file_cond_queue():
> =C2=A0 * In nfsd_file_put(), nf->nf_ref is 1, so it tries to do nfsd_file=
_lru_add().
> =C2=A0 * nfsd_file_lru_add() sets REFERENCED bit and returns true.
> =C2=A0 * Some userland application runs 'exportfs -f' or something like
> that, which triggers
> =C2=A0=C2=A0=C2=A0 __nfsd_file_cache_purge() -> nfsd_file_cond_queue().
> =C2=A0 * In nfsd_file_cond_queue(), it runs [if (!nfsd_file_unhash(nf))],
> unhash is done
> =C2=A0=C2=A0=C2=A0 successfully.
> =C2=A0 * nfsd_file_cond_queue() runs [if (!nfsd_file_get(nf))], now
> nf->nf_ref goes to 2.
> =C2=A0 * nfsd_file_cond_queue() runs [if (nfsd_file_lru_remove(nf))], it =
succeeds.
> =C2=A0 * nfsd_file_cond_queue() runs [if (refcount_sub_and_test(decrement=
,
> &nf->nf_ref))]
> =C2=A0=C2=A0=C2=A0 (with "decrement" being 2), so the nf->nf_ref goes to =
0, the 'nf'
> is added to the
> =C2=A0=C2=A0=C2=A0 dispose list [list_add(&nf->nf_lru, dispose)]
> =C2=A0 * nfsd_file_put() detects NFSD_FILE_HASHED bit is cleared, so it
> tries to remove
> =C2=A0=C2=A0=C2=A0 the 'nf' from the LRU [if (!nfsd_file_lru_remove(nf))]=
, although
> the 'nf' is not
> =C2=A0=C2=A0=C2=A0 in the LRU, but it is linked in the 'dispose' list,
> nfsd_file_lru_remove() simply
> =C2=A0=C2=A0=C2=A0 treats it as part of the LRU and removes it. This lead=
s to its removal from
> =C2=A0=C2=A0=C2=A0 the 'dispose' list!
> =C2=A0 * Now nf->ref is 0, unhashed. nfsd_file_put() continues its execut=
ion and set
> =C2=A0=C2=A0=C2=A0 nf->nf_ref to REFCOUNT_SATURATED.
>=20
> As shown in the above analysis, using nf_lru for both the LRU list and
> dispose list
> can cause the leaks. This patch adds a new list_head nf_gc in struct
> nfsd_file, and uses
> it for the dispose list. It's not expected to have a nfsd_file
> unhashed but it's not
> added to the dispose list, so in nfsd_file_cond_queue() and
> nfsd_file_lru_cb() nfsd_file
> is unhashed after being added to the dispose list.
>=20

I don't see where we require the object to be either hashed or on the
dispose list.  I think you probably just want to do a patch that
changes the dispose list to use a dedicated list_head without
reordering when the these things are unhashed.

> Signed-off-by: Youzhong Yang <youzhong@gmail.com>
> ---
> =C2=A0fs/nfsd/filecache.c | 23 ++++++++++++++---------
> =C2=A0fs/nfsd/filecache.h |=C2=A0 1 +
> =C2=A02 files changed, 15 insertions(+), 9 deletions(-)
>=20
> diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> index ad9083ca144b..3aef2ddfce94 100644
> --- a/fs/nfsd/filecache.c
> +++ b/fs/nfsd/filecache.c
> @@ -216,6 +216,7 @@ nfsd_file_alloc(struct net *net, struct inode
> *inode, unsigned char need,
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 return NULL;
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 INIT_LIST_HEAD(&nf->nf_lru);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 INIT_LIST_HEAD(&nf->nf_gc);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nf->nf_birthtime =3D ktime_get=
();
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nf->nf_file =3D NULL;
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nf->nf_cred =3D get_current_cr=
ed();
> @@ -393,8 +394,8 @@ nfsd_file_dispose_list(struct list_head *dispose)
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct nfsd_file *nf;
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 while (!list_empty(dispose)) {
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 nf =3D list_first_entry(dispose, struct nfsd_file, nf_lru);
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_del_init(&nf->nf_lru);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 nf =3D list_first_entry(dispose, struct nfsd_file, nf_gc);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_del_init(&nf->nf_gc);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 nfsd_file_free(nf);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 }
> =C2=A0}
> @@ -411,12 +412,12 @@ nfsd_file_dispose_list_delayed(struct list_head *di=
spose)
> =C2=A0{
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 while(!list_empty(dispose)) {
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 struct nfsd_file *nf =3D list_first_entry(dispose,
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct nfsd_file, nf=
_lru);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct nfsd_file, nf=
_gc);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 struct nfsd_net *nn =3D net_generic(nf->nf_net, nfsd_net=
_id);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 struct nfsd_fcache_disposal *l =3D nn->fcache_disposal;
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 spin_lock(&l->lock);
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_move_tail(&nf->nf_lru, &l->freeme);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_move_tail(&nf->nf_gc, &l->freeme);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 spin_unlock(&l->lock);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 svc_wake_up(nn->nfsd_serv);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 }
> @@ -502,8 +503,10 @@ nfsd_file_lru_cb(struct list_head *item, struct
> list_lru_one *lru,
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 }
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* Refcount went to zero. Unha=
sh it and queue it to the dispose list */
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 list_lru_isolate(lru, &nf->nf_lru);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 list_add(&nf->nf_gc, head);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* Unhash after removing from LRU a=
nd adding to dispose list */
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nfsd_file_unhash(nf);
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 list_lru_isolate_move(lru, &nf->nf_=
lru, head);

I don't see the point in reordering these operations. Hashing is all
about making the thing findable by nfsd operations. The _last_ thing we
want to do is put it on the dispose list while the thing can still be
found by nfsd threads doing operations.

> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 this_cpu_inc(nfsd_file_evictio=
ns);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 trace_nfsd_file_gc_disposed(nf=
);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 return LRU_REMOVED;
> @@ -565,7 +568,7 @@ nfsd_file_cond_queue(struct nfsd_file *nf, struct
> list_head *dispose)
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 int decrement =3D 1;
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* If we raced with someone el=
se unhashing, ignore it */
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (!nfsd_file_unhash(nf))
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (!test_bit(NFSD_FILE_HASHED, &nf=
->nf_flags))

The above change looks wrong. I don't think we need to change this.

> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 return;
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* If we can't get a reference=
, ignore it */
> @@ -578,7 +581,9 @@ nfsd_file_cond_queue(struct nfsd_file *nf, struct
> list_head *dispose)
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* If refcount goes to 0, then=
 put on the dispose list */
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (refcount_sub_and_test(decr=
ement, &nf->nf_ref)) {
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_add(&nf->nf_lru, dispose);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_add(&nf->nf_gc, dispose);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 /* Unhash after adding to dispose list */
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 nfsd_file_unhash(nf);

This too looks wrong? Maybe I'm unclear on the race you're trying to
fix with this? What's the harm in unhashing it early?

> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 trace_nfsd_file_closing(nf);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 }
> =C2=A0}
> @@ -654,8 +659,8 @@ nfsd_file_close_inode_sync(struct inode *inode)
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nfsd_file_queue_for_close(inod=
e, &dispose);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 while (!list_empty(&dispose)) =
{
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 nf =3D list_first_entry(&dispose, struct nfsd_file, nf_lru);
> -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_del_init(&nf->nf_lru);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 nf =3D list_first_entry(&dispose, struct nfsd_file, nf_gc);
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0 list_del_init(&nf->nf_gc);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 nfsd_file_free(nf);
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 }
> =C2=A0}
> diff --git a/fs/nfsd/filecache.h b/fs/nfsd/filecache.h
> index c61884def906..3fbec24eea6c 100644
> --- a/fs/nfsd/filecache.h
> +++ b/fs/nfsd/filecache.h
> @@ -44,6 +44,7 @@ struct nfsd_file {
>=20
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct nfsd_file_mark=C2=A0=C2=
=A0 *nf_mark;
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct list_head=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nf_lru;
> +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct list_head=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0 nf_gc;
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 struct rcu_head=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nf_rcu;
> =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 ktime_t=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
 nf_birthtime;
> =C2=A0};
> --
> 2.34.1
>
> On Thu, Jul 4, 2024 at 7:14=E2=80=AFAM Jeff Layton <jlayton@poochiereds.n=
et> wrote:
> >=20
> > On Wed, 2024-07-03 at 16:46 -0400, Youzhong Yang wrote:
> > > Thank you Chuck. Here are my quick answers to your comments:
> > >=20
> > > - I don't have a quick reproducer. I reproduced it by using hundreds
> > > of nfs clients generating +600K ops under our workload in the testing
> > > environment. Theoretically it should be possible to simplify the
> > > reproduction but I am still working on it.
> > >=20
> > > -=C2=A0 I understand zfs is an out-of-tree file system. That's fine. =
But
> > > this leaking can happen to any file system, and leaking is not a good
> > > thing no matter what file system it is.
> > >=20
> > > -=C2=A0 I will try to come up with a reproducer using xfs or btrfs if=
 possible.
> > >=20
> > > Now back to the problem itself, here are my findings:
> > >=20
> > > - nfsd_file_put() in one thread can race with another thread doing
> > > garbage collection (running nfsd_file_gc() -> list_lru_walk() ->
> > > nfsd_file_lru_cb()):
> > >=20
> > > =C2=A0 * In nfsd_file_put(), nf->nf_ref is 1, so it tries to do nfsd_=
file_lru_add().
> > > =C2=A0 * nfsd_file_lru_add() returns true (thus NFSD_FILE_REFERENCED =
bit
> > > set for nf->nf_flags)
> > > =C2=A0 * garbage collector kicks in, nfsd_file_lru_cb() clears REFERE=
NCED
> > > bit and returns LRU_ROTATE.
> > > =C2=A0 * garbage collector kicks in again, nfsd_file_lru_cb() now
> > > decrements nf->nf_ref to 0, runs nfsd_file_unhash(), removes it from
> > > the LRU and adds to the dispose list [list_lru_isolate_move(lru,
> > > &nf->nf_lru, head);]
> > > =C2=A0 * nfsd_file_put() detects NFSD_FILE_HASHED bit is cleared, so =
it
> > > tries to remove the 'nf' from the LRU [if (!nfsd_file_lru_remove(nf))=
]
> > > =C2=A0 * The 'nf' has been added to the 'dispose' list by
> > > nfsd_file_lru_cb(), so nfsd_file_lru_remove(nf) simply treats it as
> > > part of the LRU and removes it, which leads it to be removed from the
> > > 'dispose' list.
> > > =C2=A0 * At this moment, nf->nf_ref is 0, it's unhashed, and not on t=
he
> > > LRU. nfsd_file_put() continues its execution [if
> > > (refcount_dec_and_test(&nf->nf_ref))], as nf->nf_ref is already 0, no=
w
> > > bad thing happens: nf->nf_ref is set to REFCOUNT_SATURATED, and the
> > > 'nf' is leaked.
> > >=20
> > > To make this happen, the right timing is crucial. It can be reproduce=
d
> > > by adding artifical delays in filecache.c, or hammering the nfsd with
> > > tons of ops.
> > >=20
> > > - Let's see how nfsd_file_put() can race with nfsd_file_cond_queue():
> > > =C2=A0 * In nfsd_file_put(), nf->nf_ref is 1, so it tries to do nfsd_=
file_lru_add().
> > > =C2=A0 * nfsd_file_lru_add() sets REFERENCED bit and returns true.
> > > =C2=A0 * 'exportfs -f' or something like that triggers
> > > __nfsd_file_cache_purge() -> nfsd_file_cond_queue().
> > > =C2=A0 * In nfsd_file_cond_queue(), it runs [if (!nfsd_file_unhash(nf=
))],
> > > unhash is done successfully.
> > > =C2=A0 * nfsd_file_cond_queue() runs [if (!nfsd_file_get(nf))], now
> > > nf->nf_ref goes to 2.
> > > =C2=A0 * nfsd_file_cond_queue() runs [if (nfsd_file_lru_remove(nf))],=
 it succeeds.
> > > =C2=A0 * nfsd_file_cond_queue() runs [if (refcount_sub_and_test(decre=
ment,
> > > &nf->nf_ref))] (with "decrement" being 2), so the nf->nf_ref goes to
> > > 0, the 'nf' is added to the dispost list [list_add(&nf->nf_lru,
> > > dispose)]
> > > =C2=A0 * nfsd_file_put() detects NFSD_FILE_HASHED bit is cleared, so =
it
> > > tries to remove the 'nf' from the LRU [if
> > > (!nfsd_file_lru_remove(nf))], although the 'nf' is not in the LRU, bu=
t
> > > it is linked in the 'dispose' list, nfsd_file_lru_remove() simply
> > > treats it as part of the LRU and removes it. This leads to its remova=
l
> > > from the 'dispose' list!
> > > =C2=A0 * Now nf->ref is 0, unhashed. nfsd_file_put() continues its
> > > execution and sets nf->nf_ref to REFCOUNT_SATURATED.
> > >=20
> > > The purpose of nf->nf_lru is problematic. As you can see, it is used
> > > for the LRU list, and also the 'dispose' list. Adding another 'struct
> > > list_head' specifically for the 'dispose' list seems to be a better
> > > way of fixing this race condition. Either way works for me.
> > >=20
> > > Would you agree my above analysis makes sense? Thanks.
> > >=20
> >=20
> > I think so. It's been a while since I've done much work in this code,
> > but it does sound like there is a race in the LRU handling.
> >=20
> >=20
> > Like Chuck said, the nf->nf_lru list should be safe to use for multiple
> > purposes, but that's only the case if we're not using that list as an
> > indicator.
> >=20
> > The list_lru code does check this:
> >=20
> > =C2=A0=C2=A0=C2=A0 if (!list_empty(item)) {
> >=20
> > ...so if we ever check this while it's sitting on the dispose list, it
> > will handle it incorrectly. It sounds like that's the root cause of the
> > problem you're seeing?
> >=20
> > If so, then maybe a separate list_head for disposal would be better.
> >=20
> > > Here is my patch with signed-off-by:
> > >=20
> > > From: Youzhong Yang <youzhong@gmail.com>
> > > Date: Mon, 1 Jul 2024 06:45:22 -0400
> > > Subject: [PATCH] nfsd: fix nfsd_file leaking due to race condition an=
d early
> > > =C2=A0unhash
> > >=20
> > > Signed-off-by: Youzhong Yang <youzhong@gmail.com>
> > > ---
> > > =C2=A0fs/nfsd/filecache.c | 14 +++++++++++++-
> > > =C2=A01 file changed, 13 insertions(+), 1 deletion(-)
> > >=20
> > > diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> > > index 1a6d5d000b85..2323829f7208 100644
> > > --- a/fs/nfsd/filecache.c
> > > +++ b/fs/nfsd/filecache.c
> > > @@ -389,6 +389,17 @@ nfsd_file_put(struct nfsd_file *nf)
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (!=
nfsd_file_lru_remove(nf))
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 return;
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0 }
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 /*
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0 * Racing with nfsd_file_cond_queue() or nfsd_file_=
lru_cb(),
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0 * it's unhashed but then removed from the dispose =
list,
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0 * so we need to free it.
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0 */
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 if (refcount_read(&nf->nf_ref) =3D=3D 0 &&
> >=20
> > A refcount_read in this path is a red flag to me. Anytime you're just
> > looking at the refcount without changing anything just screams out
> > "race condition".
> >=20
> > In this case, what guarantee is there that this won't run afoul of the
> > timing? We could check this and find out it's 1 just before it goes to
> > 0 and you check the other conditions.
> >=20
> > Does anything prevent that?
> >=20
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 !test_bit(NFSD_FILE_HASHED, &nf-=
>nf_flags) &&
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 list_empty(&nf->nf_lru)) {
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 nfsd_fil=
e_free(nf);
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 return;
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 }
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 }
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (refcount_dec_and_test(=
&nf->nf_ref))
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0 nfsd_file_free(nf);
> > > @@ -576,7 +587,7 @@ nfsd_file_cond_queue(struct nfsd_file *nf, struct
> > > list_head *dispose)
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 int decrement =3D 1;
> > >=20
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* If we raced with someon=
e else unhashing, ignore it */
> > > -=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (!nfsd_file_unhash(nf))
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (!test_bit(NFSD_FILE_HASHED,=
 &nf->nf_flags))
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0 return;
> >=20
> > Same here: you're just testing for the HASHED bit, but could this also
> > race with someone who is setting it just after you get here. Why is
> > that not a problem?
> >=20
> > >=20
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* If we can't get a refer=
ence, ignore it */
> > > @@ -590,6 +601,7 @@ nfsd_file_cond_queue(struct nfsd_file *nf, struct
> > > list_head *dispose)
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 /* If refcount goes to 0, =
then put on the dispose list */
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 if (refcount_sub_and_test(=
decrement, &nf->nf_ref)) {
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0 list_add(&nf->nf_lru, dispose);
> > > +=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=
=C2=A0=C2=A0=C2=A0 nfsd_file_unhash(nf);
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=
=A0=C2=A0=C2=A0=C2=A0 trace_nfsd_file_closing(nf);
> > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 }
> > > =C2=A0}
> > > --
> > > 2.43.0
> > >=20
> > > On Wed, Jul 3, 2024 at 2:21=E2=80=AFPM Chuck Lever <chuck.lever@oracl=
e.com> wrote:
> > > >=20
> > > > On Wed, Jul 03, 2024 at 10:12:33AM -0400, Youzhong Yang wrote:
> > > > > Hello,
> > > > >=20
> > > > > I'd like to report a nfsd_file leaking issue and propose a fix fo=
r it.
> > > > >=20
> > > > > When I tested Linux kernel 6.8 and 6.6, I noticed nfsd_file leaks
> > > > > which led to undestroyable file systems (zfs),
> > > >=20
> > > > Thanks for the report. Some initial comments:
> > > >=20
> > > > - Do you have a specific reproducer? In other words, what is the
> > > > =C2=A0 simplest program that can run on an NFS client that will tri=
gger
> > > > =C2=A0 this leak, and can you post it?
> > > >=20
> > > > - "zfs" is an out-of-tree file system, so it's not directly
> > > > =C2=A0 supported for NFSD.
> > > >=20
> > > > - The guidelines for patch submission require us to fix issues in
> > > > =C2=A0 upstream Linux first (currently that's v6.10-rc6). Then that=
 fix
> > > > =C2=A0 can be backported to older stable kernels like 6.6.
> > > >=20
> > > > Can you reproduce the leak with one of the in-kernel filesystems
> > > > (either xfs or btrfs would be great) and with NFSD in 6.10-rc6?
> > > >=20
> > > > One more comment below.
> > > >=20
> > > >=20
> > > > > here are some examples:
> > > > >=20
> > > > > crash> struct nfsd_file -x ffff88e160db0460
> > > > > struct nfsd_file {
> > > > > =C2=A0 nf_rlist =3D {
> > > > > =C2=A0=C2=A0=C2=A0 rhead =3D {
> > > > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 next =3D 0xffff8921fa2392f1
> > > > > =C2=A0=C2=A0=C2=A0 },
> > > > > =C2=A0=C2=A0=C2=A0 next =3D 0x0
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_inode =3D 0xffff8882bc312ef8,
> > > > > =C2=A0 nf_file =3D 0xffff88e2015b1500,
> > > > > =C2=A0 nf_cred =3D 0xffff88e3ab0e7800,
> > > > > =C2=A0 nf_net =3D 0xffffffff83d41600 <init_net>,
> > > > > =C2=A0 nf_flags =3D 0x8,
> > > > > =C2=A0 nf_ref =3D {
> > > > > =C2=A0=C2=A0=C2=A0 refs =3D {
> > > > > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 counter =3D 0xc0000000
> > > > > =C2=A0=C2=A0=C2=A0 }
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_may =3D 0x4,
> > > > > =C2=A0 nf_mark =3D 0xffff88e1bddfb320,
> > > > > =C2=A0 nf_lru =3D {
> > > > > =C2=A0=C2=A0=C2=A0 next =3D 0xffff88e160db04a8,
> > > > > =C2=A0=C2=A0=C2=A0 prev =3D 0xffff88e160db04a8
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_rcu =3D {
> > > > > =C2=A0=C2=A0=C2=A0 next =3D 0x10000000000,
> > > > > =C2=A0=C2=A0=C2=A0 func =3D 0x0
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_birthtime =3D 0x73d22fc1728
> > > > > }
> > > > >=20
> > > > > crash> struct nfsd_file.nf_flags,nf_ref.refs.counter,nf_lru,nf_fi=
le -x
> > > > > ffff88839a53d850
> > > > > =C2=A0 nf_flags =3D 0x8,
> > > > > =C2=A0 nf_ref.refs.counter =3D 0x0
> > > > > =C2=A0 nf_lru =3D {
> > > > > =C2=A0=C2=A0=C2=A0 next =3D 0xffff88839a53d898,
> > > > > =C2=A0=C2=A0=C2=A0 prev =3D 0xffff88839a53d898
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_file =3D 0xffff88810ede8700,
> > > > >=20
> > > > > crash> struct nfsd_file.nf_flags,nf_ref.refs.counter,nf_lru,nf_fi=
le -x
> > > > > ffff88c32b11e850
> > > > > =C2=A0 nf_flags =3D 0x8,
> > > > > =C2=A0 nf_ref.refs.counter =3D 0x0
> > > > > =C2=A0 nf_lru =3D {
> > > > > =C2=A0=C2=A0=C2=A0 next =3D 0xffff88c32b11e898,
> > > > > =C2=A0=C2=A0=C2=A0 prev =3D 0xffff88c32b11e898
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_file =3D 0xffff88c20a701c00,
> > > > >=20
> > > > > crash> struct nfsd_file.nf_flags,nf_ref.refs.counter,nf_lru,nf_fi=
le -x
> > > > > ffff88e372709700
> > > > > =C2=A0 nf_flags =3D 0xc,
> > > > > =C2=A0 nf_ref.refs.counter =3D 0x0
> > > > > =C2=A0 nf_lru =3D {
> > > > > =C2=A0=C2=A0=C2=A0 next =3D 0xffff88e372709748,
> > > > > =C2=A0=C2=A0=C2=A0 prev =3D 0xffff88e372709748
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_file =3D 0xffff88e0725e6400,
> > > > >=20
> > > > > crash> struct nfsd_file.nf_flags,nf_ref.refs.counter,nf_lru,nf_fi=
le -x
> > > > > ffff8982864944d0
> > > > > =C2=A0 nf_flags =3D 0xc,
> > > > > =C2=A0 nf_ref.refs.counter =3D 0x0
> > > > > =C2=A0 nf_lru =3D {
> > > > > =C2=A0=C2=A0=C2=A0 next =3D 0xffff898286494518,
> > > > > =C2=A0=C2=A0=C2=A0 prev =3D 0xffff898286494518
> > > > > =C2=A0 },
> > > > > =C2=A0 nf_file =3D 0xffff89803c0ff700,
> > > > >=20
> > > > > The leak occurs when nfsd_file_put() races with nfsd_file_cond_qu=
eue()
> > > > > or nfsd_file_lru_cb(). With the following patch, I haven't observ=
ed
> > > > > any leak after a few days heavy nfs load:
> > > >=20
> > > > Our patch submission guidelines require a Signed-off-by:
> > > > line at the end of the patch description. See the "Sign your work -
> > > > the Developer's Certificate of Origin" section of
> > > >=20
> > > > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/=
tree/Documentation/process/submitting-patches.rst?h=3Dv6.10-rc6
> > > >=20
> > > > (Needed here in case your fix is acceptable).
> > > >=20
> > > >=20
> > > > > diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> > > > > index 1a6d5d000b85..2323829f7208 100644
> > > > > --- a/fs/nfsd/filecache.c
> > > > > +++ b/fs/nfsd/filecache.c
> > > > > @@ -389,6 +389,17 @@ nfsd_file_put(struct nfsd_file *nf)
> > > > > =C2=A0 if (!nfsd_file_lru_remove(nf))
> > > > > =C2=A0 return;
> > > > > =C2=A0 }
> > > > > + /*
> > > > > + * Racing with nfsd_file_cond_queue() or nfsd_file_lru_cb(),
> > > > > + * it's unhashed but then removed from the dispose list,
> > > > > + * so we need to free it.
> > > > > + */
> > > > > + if (refcount_read(&nf->nf_ref) =3D=3D 0 &&
> > > > > +=C2=A0=C2=A0=C2=A0=C2=A0 !test_bit(NFSD_FILE_HASHED, &nf->nf_fla=
gs) &&
> > > > > +=C2=A0=C2=A0=C2=A0=C2=A0 list_empty(&nf->nf_lru)) {
> > > > > + nfsd_file_free(nf);
> > > > > + return;
> > > > > + }
> > > > > =C2=A0 }
> > > > > =C2=A0 if (refcount_dec_and_test(&nf->nf_ref))
> > > > > =C2=A0 nfsd_file_free(nf);
> > > > > @@ -576,7 +587,7 @@ nfsd_file_cond_queue(struct nfsd_file *nf, st=
ruct
> > > > > list_head *dispose)
> > > > > =C2=A0 int decrement =3D 1;
> > > > >=20
> > > > > =C2=A0 /* If we raced with someone else unhashing, ignore it */
> > > > > - if (!nfsd_file_unhash(nf))
> > > > > + if (!test_bit(NFSD_FILE_HASHED, &nf->nf_flags))
> > > > > =C2=A0 return;
> > > > >=20
> > > > > =C2=A0 /* If we can't get a reference, ignore it */
> > > > > @@ -590,6 +601,7 @@ nfsd_file_cond_queue(struct nfsd_file *nf, st=
ruct
> > > > > list_head *dispose)
> > > > > =C2=A0 /* If refcount goes to 0, then put on the dispose list */
> > > > > =C2=A0 if (refcount_sub_and_test(decrement, &nf->nf_ref)) {
> > > > > =C2=A0 list_add(&nf->nf_lru, dispose);
> > > > > + nfsd_file_unhash(nf);
> > > > > =C2=A0 trace_nfsd_file_closing(nf);
> > > > > =C2=A0 }
> > > > > =C2=A0}
> > > > >=20
> > > > > Please kindly review the patch and let me know if it makes sense.
> > > > >=20
> > > > > Thanks,
> > > > >=20
> > > > > -Youzhong
> > > > >=20
> > > >=20
> > > > --
> > > > Chuck Lever
> > >=20
> >=20
> > --
> > Jeff Layton <jlayton@poochiereds.net>

--=20
Jeff Layton <jlayton@poochiereds.net>

--=20
Jeff Layton <jlayton@kernel.org>