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;
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
In-Reply-To: <20220510142247.16071-1-wanjiabing@vivo.com>
References: <20220510142247.16071-1-wanjiabing@vivo.com>
From:   Antoine Tenart <atenart@kernel.org>
To:     "David S. Miller" <davem@davemloft.net>,
        Andrew Lunn <andrew@lunn.ch>,
        Eric Dumazet <edumazet@google.com>,
        Heiner Kallweit <hkallweit1@gmail.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        Russell King <linux@armlinux.org.uk>,
        Wan Jiabing <wanjiabing@vivo.com>,
        linux-kernel@vger.kernel.org, netdev@vger.kernel.org
Subject: Re: [PATCH v2 net] net: phy: mscc: Add error check when __phy_read() failed
Message-ID: <165219411356.3924.11722336879963021691@kwain>
Date:   Tue, 10 May 2022 16:48:33 +0200
Precedence: bulk

Hello,

Quoting Wan Jiabing (2022-05-10 16:22:45)
> Calling __phy_read() might return a negative error code. Use 'int'
> to declare variables which call __phy_read() and also add error check
> for them.
>=20
> The numerous callers of vsc8584_macsec_phy_read() don't expect it to
> fail. So don't return the error code from __phy_read(), but also don't
> return random values if it does fail.
>=20
> Fixes: fa164e40c53b ("net: phy: mscc: split the driver into separate file=
s")

Does this fix an actual issue or was this found by code inspection? If
that is not fixing a real issue I don't think it should go to stable
trees.

Also this is not the right commit, the __phy_read call was introduced
before splitting the file.

>  static u32 vsc8584_macsec_phy_read(struct phy_device *phydev,
>                                    enum macsec_bank bank, u32 reg)
>  {
> -       u32 val, val_l =3D 0, val_h =3D 0;
> +       int rc, val, val_l, val_h;
>         unsigned long deadline;
> -       int rc;
> +       u32 ret =3D 0;
> =20
>         rc =3D phy_select_page(phydev, MSCC_PHY_PAGE_MACSEC);
>         if (rc < 0)
> @@ -47,15 +47,20 @@ static u32 vsc8584_macsec_phy_read(struct phy_device =
*phydev,
>         deadline =3D jiffies + msecs_to_jiffies(PROC_CMD_NCOMPLETED_TIMEO=
UT_MS);
>         do {
>                 val =3D __phy_read(phydev, MSCC_EXT_PAGE_MACSEC_19);
> +               if (val < 0)
> +                       goto failed;
>         } while (time_before(jiffies, deadline) && !(val & MSCC_PHY_MACSE=
C_19_CMD));
> =20
>         val_l =3D __phy_read(phydev, MSCC_EXT_PAGE_MACSEC_17);
>         val_h =3D __phy_read(phydev, MSCC_EXT_PAGE_MACSEC_18);
> =20
> +       if (val_l > 0 && val_h > 0)
> +               ret =3D (val_h << 16) | val_l;

Both values have to be non-0 for the function to return a value? I
haven't checked but I would assume it is valid to have one of the two
being 0.

>  failed:
>         phy_restore_page(phydev, rc, rc);
> =20
> -       return (val_h << 16) | val_l;
> +       return ret;
>  }

Thanks,
Antoine