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;
Message-ID: <e94f4967a770b7004686155291c885e61a3403b2.camel@gmail.com>
Subject: Re: [PATCH v2 net-next 2/3] net: dsa: mv88e6xxx: disable hold of
 chip lock for handling
From:   Alexander H Duyck <alexander.duyck@gmail.com>
To:     "Hans J. Schultz" <netdev@kapio-technology.com>,
        davem@davemloft.net, kuba@kernel.org
Cc:     netdev@vger.kernel.org, Andrew Lunn <andrew@lunn.ch>,
        Florian Fainelli <f.fainelli@gmail.com>,
        Vladimir Oltean <olteanv@gmail.com>,
        Eric Dumazet <edumazet@google.com>,
        Paolo Abeni <pabeni@redhat.com>,
        open list <linux-kernel@vger.kernel.org>
Date:   Tue, 13 Dec 2022 14:06:33 -0800
In-Reply-To: <20221213174650.670767-3-netdev@kapio-technology.com>
References: <20221213174650.670767-1-netdev@kapio-technology.com>
         <20221213174650.670767-3-netdev@kapio-technology.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4 (3.44.4-2.fc36) 
MIME-Version: 1.0
Precedence: bulk

On Tue, 2022-12-13 at 18:46 +0100, Hans J. Schultz wrote:
> As functions called under the interrupt handler will need to take the
> netlink lock, we need to release the chip lock before calling those
> functions as otherwise double lock deadlocks will occur as userspace
> calls towards the driver often take the netlink lock and then the
> chip lock.
>=20
> The deadlock would look like:
>=20
> Interrupt handler: chip lock taken, but cannot take netlink lock as
>                    userspace config call has netlink lock.
> Userspace config: netlink lock taken, but cannot take chip lock as
>                    the interrupt handler has the chip lock.
>=20
> Signed-off-by: Hans J. Schultz <netdev@kapio-technology.com>

Just to confirm, in order to see the deadlocks I would imagine you need
something in the tracepoints that is taking the netlink lock?

If so you might want to reference the commit that switched out the
dev_err_ratelimited for the tracepoints 8646384d80f3 ("net: dsa:
mv88e6xxx: replace ATU violation prints with trace points") so that
this patch can be found an applied to any kernels that pull in those
tracepoints.

> ---
>  drivers/net/dsa/mv88e6xxx/global1_atu.c | 14 ++++++++------
>  1 file changed, 8 insertions(+), 6 deletions(-)
>=20
> diff --git a/drivers/net/dsa/mv88e6xxx/global1_atu.c b/drivers/net/dsa/mv=
88e6xxx/global1_atu.c
> index 61ae2d61e25c..34203e112eef 100644
> --- a/drivers/net/dsa/mv88e6xxx/global1_atu.c
> +++ b/drivers/net/dsa/mv88e6xxx/global1_atu.c
> @@ -409,11 +409,11 @@ static irqreturn_t mv88e6xxx_g1_atu_prob_irq_thread=
_fn(int irq, void *dev_id)
> =20
>  	err =3D mv88e6xxx_g1_read_atu_violation(chip);
>  	if (err)
> -		goto out;
> +		goto out_unlock;
> =20
>  	err =3D mv88e6xxx_g1_read(chip, MV88E6XXX_G1_ATU_OP, &val);
>  	if (err)
> -		goto out;
> +		goto out_unlock;
> =20
>  	err =3D mv88e6xxx_g1_atu_fid_read(chip, &fid);
>  	if (err)
> @@ -421,11 +421,13 @@ static irqreturn_t mv88e6xxx_g1_atu_prob_irq_thread=
_fn(int irq, void *dev_id)
> =20
>  	err =3D mv88e6xxx_g1_atu_data_read(chip, &entry);
>  	if (err)
> -		goto out;
> +		goto out_unlock;
> =20
>  	err =3D mv88e6xxx_g1_atu_mac_read(chip, &entry);
>  	if (err)
> -		goto out;
> +		goto out_unlock;
> +
> +	mv88e6xxx_reg_unlock(chip);
> =20
>  	spid =3D entry.state;
> =20
> @@ -449,13 +451,13 @@ static irqreturn_t mv88e6xxx_g1_atu_prob_irq_thread=
_fn(int irq, void *dev_id)
>  						   fid);
>  		chip->ports[spid].atu_full_violation++;
>  	}
> -	mv88e6xxx_reg_unlock(chip);
> =20
>  	return IRQ_HANDLED;
> =20
> -out:
> +out_unlock:
>  	mv88e6xxx_reg_unlock(chip);
> =20
> +out:
>  	dev_err(chip->dev, "ATU problem: error %d while handling interrupt\n",
>  		err);
>  	return IRQ_HANDLED;

Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>