Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <3e819d22808f7f1f232bc1242f2260106f37f875.camel@sipsolutions.net>
Subject: Re: [PATCH 1/2] cfg80211: fix dead lock for nl80211_new_interface()
From:   Johannes Berg <johannes@sipsolutions.net>
To:     Aran Dalton <arda@allwinnertech.com>, davem@davemloft.net,
        edumazet@google.com, kuba@kernel.org, pabeni@redhat.com
Cc:     linux-wireless@vger.kernel.org, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org
Date:   Wed, 21 Sep 2022 18:36:18 +0200
In-Reply-To: <20220921091913.110749-1-arda@allwinnertech.com>
References: <20220921091913.110749-1-arda@allwinnertech.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4 (3.44.4-1.fc36) 
MIME-Version: 1.0
Precedence: bulk

On Wed, 2022-09-21 at 17:19 +0800, Aran Dalton wrote:
> Both nl80211_new_interface and cfg80211_netdev_notifier_call hold the
> same wiphy_lock, then cause deadlock.
>=20
> The main call stack as bellow:
>=20
> nl80211_new_interface() takes wiphy_lock
>  -> _nl80211_new_interface:
>   -> rdev_add_virtual_intf
>    -> rdev->ops->add_virtual_intf
>     -> register_netdevice

The bug is yours, here, you're no longer allowed to call
register_netdevice() here.

If you have an out-of-tree driver that we couldn't update when doing
tree-wide changes, you probably shouldn't assume that the bug is
upstream and send random locking patches ... :)

johannes