Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   Qian Cai <cai@lca.pw>
Mime-Version: 1.0 (1.0)
Subject: Re: [PATCH] skbuff: fix a data race in skb_queue_len()
Date:   Mon, 3 Feb 2020 17:34:34 -0500
Message-Id: <A5748E52-90D8-4B5B-87BD-65F980D79A63@lca.pw>
References: <126aa227-2383-efdd-742c-6ccf85ae7ba4@gmail.com>
Cc:     davem@davemloft.net, kuba@kernel.org, elver@google.com,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org
In-Reply-To: <126aa227-2383-efdd-742c-6ccf85ae7ba4@gmail.com>
To:     Eric Dumazet <eric.dumazet@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


> On Feb 3, 2020, at 3:28 PM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
>=20
> We added recently skb_queue_empty_lockless() helper, to use in these conte=
xts.
>=20
> The fact that we use READ_ONCE() is more of an implementation detail I thi=
nk.
>=20

Make sense. I=E2=80=99ll use lockless in naming instead.

> Also, addressing load-stearing issues without making sure the write side
> is using WRITE_ONCE() might be not enough (even if KCSAN warnings disappea=
r)

I suppose that could be a case. I=E2=80=99ll have,

WRITE_ONCE(list->qlen, list->qlen - 1);

in __skb_unlink() where it had already had a few WRITE_ONCE() for other vari=
ables.=