Received-SPF: pass (google.com: domain of linux-bluetooth-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220526094918.482971-1-niejianglei2021@163.com> <081b216e6496e8cc2284df191dcc2d8b604d04f7.camel@redhat.com>
In-Reply-To: <081b216e6496e8cc2284df191dcc2d8b604d04f7.camel@redhat.com>
From:   Luiz Augusto von Dentz <luiz.dentz@gmail.com>
Date:   Thu, 26 May 2022 13:11:56 -0700
Message-ID: <CABBYNZJ+E6KMyqODib_nhGJuZzWssSrswKR9MoqPrM7tuEpDcg@mail.gmail.com>
Subject: Re: [PATCH] Bluetooth: hci_conn: fix potential double free in le_scan_cleanup()
To:     Paolo Abeni <pabeni@redhat.com>
Cc:     Jianglei Nie <niejianglei2021@163.com>, marcel@holtmann.org,
        johan.hedberg@gmail.com, davem@davemloft.net, edumazet@google.com,
        kuba@kernel.org, linux-bluetooth@vger.kernel.org,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi,

On Thu, May 26, 2022 at 4:24 AM Paolo Abeni <pabeni@redhat.com> wrote:
>
> On Thu, 2022-05-26 at 17:49 +0800, Jianglei Nie wrote:
> > When "c == conn" is true, hci_conn_cleanup() is called. The
> > hci_conn_cleanup() calls hci_dev_put() and hci_conn_put() in
> > its function implementation. hci_dev_put() and hci_conn_put()
> > will free the relevant resource if the reference count reaches
> > zero, which may lead to a double free when hci_dev_put() and
> > hci_conn_put() are called again.
> >
> > We should add a return to this function after hci_conn_cleanup()
> > is called.
> >
> > Signed-off-by: Jianglei Nie <niejianglei2021@163.com>
> > ---
> >  net/bluetooth/hci_conn.c | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/net/bluetooth/hci_conn.c b/net/bluetooth/hci_conn.c
> > index fe803bee419a..7b3e91eb9fa3 100644
> > --- a/net/bluetooth/hci_conn.c
> > +++ b/net/bluetooth/hci_conn.c
> > @@ -166,6 +166,7 @@ static void le_scan_cleanup(struct work_struct *work)
> >       if (c == conn) {
> >               hci_connect_le_scan_cleanup(conn);
> >               hci_conn_cleanup(conn);
> > +             return;
>
> This looks not correct. At very least you should release the
> hci_dev_lock.

Yep, it should probably use break instead of return.

> Cheers,
>
> Paolo
>


-- 
Luiz Augusto von Dentz