Received-SPF: pass (google.com: domain of linux-kernel-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: <CABBYNZKSnFJkyMoHn-TU1VJQz3WNNt0pC8Nvzdxb3-4-RtcQGw@mail.gmail.com>
 <20220930160843.818893-1-iam@sung-woo.kim>
In-Reply-To: <20220930160843.818893-1-iam@sung-woo.kim>
From:   Luiz Augusto von Dentz <luiz.dentz@gmail.com>
Date:   Fri, 30 Sep 2022 10:46:41 -0700
Message-ID: <CABBYNZLyCzQ_RUJKUi8dpZorPjUsyCxXcZ-ScmMHWx0a86Ra5w@mail.gmail.com>
Subject: Re: KASAN: use-after-free in __mutex_lock
To:     Sungwoo Kim <iam@sung-woo.kim>
Cc:     davem@davemloft.net, edumazet@google.com, johan.hedberg@gmail.com,
        kuba@kernel.org, linux-bluetooth@vger.kernel.org,
        linux-kernel@vger.kernel.org, marcel@holtmann.org,
        netdev@vger.kernel.org, pabeni@redhat.com,
        syzkaller@googlegroups.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Kim,

On Fri, Sep 30, 2022 at 9:09 AM Sungwoo Kim <iam@sung-woo.kim> wrote:
>
> Hi Dentz,
> How about to use l2cap_get_chan_by_scid because it looks resposible to
> handle ref_cnt.
>
> Signed-off-by: Sungwoo Kim <iam@sung-woo.kim>
> ---
>  net/bluetooth/l2cap_core.c | 24 +++++++-----------------
>  1 file changed, 7 insertions(+), 17 deletions(-)
>
> diff --git a/net/bluetooth/l2cap_core.c b/net/bluetooth/l2cap_core.c
> index 2c9de67da..d3a074cbc 100644
> --- a/net/bluetooth/l2cap_core.c
> +++ b/net/bluetooth/l2cap_core.c
> @@ -4291,26 +4291,18 @@ static int l2cap_connect_create_rsp(struct l2cap_conn *conn,
>         BT_DBG("dcid 0x%4.4x scid 0x%4.4x result 0x%2.2x status 0x%2.2x",
>                dcid, scid, result, status);
>
> -       mutex_lock(&conn->chan_lock);
> -
>         if (scid) {
> -               chan = __l2cap_get_chan_by_scid(conn, scid);
> -               if (!chan) {
> -                       err = -EBADSLT;
> -                       goto unlock;
> -               }
> +               chan = l2cap_get_chan_by_scid(conn, scid);
> +               if (!chan)
> +                       return -EBADSLT;
>         } else {
> -               chan = __l2cap_get_chan_by_ident(conn, cmd->ident);
> -               if (!chan) {
> -                       err = -EBADSLT;
> -                       goto unlock;
> -               }
> +               chan = l2cap_get_chan_by_ident(conn, cmd->ident);
> +               if (!chan)
> +                       return -EBADSLT;
>         }
>
>         err = 0;
>
> -       l2cap_chan_lock(chan);
> -
>         switch (result) {
>         case L2CAP_CR_SUCCESS:
>                 l2cap_state_change(chan, BT_CONFIG);
> @@ -4336,9 +4328,7 @@ static int l2cap_connect_create_rsp(struct l2cap_conn *conn,
>         }
>
>         l2cap_chan_unlock(chan);
> -
> -unlock:
> -       mutex_unlock(&conn->chan_lock);
> +       l2cap_chan_put(chan);
>
>         return err;
>  }
> --
> 2.25.1

Ive sent a fix yesterday:

https://patchwork.kernel.org/project/bluetooth/patch/20220929203241.4140795-1-luiz.dentz@gmail.com/

Both are sorta similar but the one above end up causing less code
changes which might be easier to backport.

-- 
Luiz Augusto von Dentz