There is a possibility xpcs_create() returned a NULL and this is not
handled properly in the sja1105 driver.
Fixed this by using IS_ERR_ON_NULL() instead of IS_ERR().
Fixes: 3ad1d171548e ("net: dsa: sja1105: migrate to xpcs for SGMII")
Cc: Vladimir Olten <[email protected]>
Signed-off-by: Wong Vee Khee <[email protected]>
---
drivers/net/dsa/sja1105/sja1105_mdio.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/dsa/sja1105/sja1105_mdio.c b/drivers/net/dsa/sja1105/sja1105_mdio.c
index 19aea8fb76f6..2c69a759ce6e 100644
--- a/drivers/net/dsa/sja1105/sja1105_mdio.c
+++ b/drivers/net/dsa/sja1105/sja1105_mdio.c
@@ -438,7 +438,7 @@ static int sja1105_mdiobus_pcs_register(struct sja1105_private *priv)
}
xpcs = xpcs_create(mdiodev, priv->phy_mode[port]);
- if (IS_ERR(xpcs)) {
+ if (IS_ERR_OR_NULL(xpcs)) {
rc = PTR_ERR(xpcs);
goto out_pcs_free;
}
--
2.25.1
On Tue, Aug 10, 2021 at 08:25:29AM +0000, Vladimir Oltean wrote:
> Hi VK,
>
> On Tue, Aug 10, 2021 at 02:35:13PM +0800, Wong Vee Khee wrote:
> > There is a possibility xpcs_create() returned a NULL and this is not
> > handled properly in the sja1105 driver.
> >
> > Fixed this by using IS_ERR_ON_NULL() instead of IS_ERR().
> >
> > Fixes: 3ad1d171548e ("net: dsa: sja1105: migrate to xpcs for SGMII")
> > Cc: Vladimir Olten <[email protected]>
> > Signed-off-by: Wong Vee Khee <[email protected]>
> > ---
> > drivers/net/dsa/sja1105/sja1105_mdio.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/dsa/sja1105/sja1105_mdio.c b/drivers/net/dsa/sja1105/sja1105_mdio.c
> > index 19aea8fb76f6..2c69a759ce6e 100644
> > --- a/drivers/net/dsa/sja1105/sja1105_mdio.c
> > +++ b/drivers/net/dsa/sja1105/sja1105_mdio.c
> > @@ -438,7 +438,7 @@ static int sja1105_mdiobus_pcs_register(struct sja1105_private *priv)
> > }
> >
> > xpcs = xpcs_create(mdiodev, priv->phy_mode[port]);
> > - if (IS_ERR(xpcs)) {
> > + if (IS_ERR_OR_NULL(xpcs)) {
> > rc = PTR_ERR(xpcs);
> > goto out_pcs_free;
> > }
> > --
> > 2.25.1
> >
>
> I think posting to 'net' is a bit drastic for this patch. I agree it is
> an issue but it has bothered absolutely nobody so far, and if sending
> this patch to 'net' means you'll be blocked with your patches until the
> 'net'->'net-next' merge, I'm not sure it's worth it.
>
Sounds reasonable for me. I will marked V2 to target net-next.
> Anyway, I don't think that PTR_ERR(xpcs) does the right thing when
> IS_NULL(xpcs). So how about you make this change:
>
> -----------------------------[ cut here ]-----------------------------
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
> index a5d150c5f3d8..ca12bf986d4d 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
> @@ -415,7 +415,7 @@ int stmmac_xpcs_setup(struct mii_bus *bus)
> continue;
>
> xpcs = xpcs_create(mdiodev, mode);
> - if (IS_ERR_OR_NULL(xpcs)) {
> + if (IS_ERR(xpcs)) {
> mdio_device_free(mdiodev);
> continue;
> }
> diff --git a/drivers/net/pcs/pcs-xpcs.c b/drivers/net/pcs/pcs-xpcs.c
> index 63fda3fc40aa..4bd61339823c 100644
> --- a/drivers/net/pcs/pcs-xpcs.c
> +++ b/drivers/net/pcs/pcs-xpcs.c
> @@ -1089,7 +1089,7 @@ struct dw_xpcs *xpcs_create(struct mdio_device *mdiodev,
>
> xpcs = kzalloc(sizeof(*xpcs), GFP_KERNEL);
> if (!xpcs)
> - return NULL;
> + return ERR_PTR(-ENOMEM);
>
> xpcs->mdiodev = mdiodev;
>
> -----------------------------[ cut here ]-----------------------------
Hi VK,
On Tue, Aug 10, 2021 at 02:35:13PM +0800, Wong Vee Khee wrote:
> There is a possibility xpcs_create() returned a NULL and this is not
> handled properly in the sja1105 driver.
>
> Fixed this by using IS_ERR_ON_NULL() instead of IS_ERR().
>
> Fixes: 3ad1d171548e ("net: dsa: sja1105: migrate to xpcs for SGMII")
> Cc: Vladimir Olten <[email protected]>
> Signed-off-by: Wong Vee Khee <[email protected]>
> ---
> drivers/net/dsa/sja1105/sja1105_mdio.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/net/dsa/sja1105/sja1105_mdio.c b/drivers/net/dsa/sja1105/sja1105_mdio.c
> index 19aea8fb76f6..2c69a759ce6e 100644
> --- a/drivers/net/dsa/sja1105/sja1105_mdio.c
> +++ b/drivers/net/dsa/sja1105/sja1105_mdio.c
> @@ -438,7 +438,7 @@ static int sja1105_mdiobus_pcs_register(struct sja1105_private *priv)
> }
>
> xpcs = xpcs_create(mdiodev, priv->phy_mode[port]);
> - if (IS_ERR(xpcs)) {
> + if (IS_ERR_OR_NULL(xpcs)) {
> rc = PTR_ERR(xpcs);
> goto out_pcs_free;
> }
> --
> 2.25.1
>
I think posting to 'net' is a bit drastic for this patch. I agree it is
an issue but it has bothered absolutely nobody so far, and if sending
this patch to 'net' means you'll be blocked with your patches until the
'net'->'net-next' merge, I'm not sure it's worth it.
Anyway, I don't think that PTR_ERR(xpcs) does the right thing when
IS_NULL(xpcs). So how about you make this change:
-----------------------------[ cut here ]-----------------------------
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
index a5d150c5f3d8..ca12bf986d4d 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
@@ -415,7 +415,7 @@ int stmmac_xpcs_setup(struct mii_bus *bus)
continue;
xpcs = xpcs_create(mdiodev, mode);
- if (IS_ERR_OR_NULL(xpcs)) {
+ if (IS_ERR(xpcs)) {
mdio_device_free(mdiodev);
continue;
}
diff --git a/drivers/net/pcs/pcs-xpcs.c b/drivers/net/pcs/pcs-xpcs.c
index 63fda3fc40aa..4bd61339823c 100644
--- a/drivers/net/pcs/pcs-xpcs.c
+++ b/drivers/net/pcs/pcs-xpcs.c
@@ -1089,7 +1089,7 @@ struct dw_xpcs *xpcs_create(struct mdio_device *mdiodev,
xpcs = kzalloc(sizeof(*xpcs), GFP_KERNEL);
if (!xpcs)
- return NULL;
+ return ERR_PTR(-ENOMEM);
xpcs->mdiodev = mdiodev;
-----------------------------[ cut here ]-----------------------------