[PATCH net v2 0/2] smsc911x: fix issues when interface is not up yet

[PATCH net v2 0/2] smsc911x: fix issues when interface is not up yet

[Wolfram Sang]

Similar to upstream commit 7f5ebf5dae42 ("ravb: avoid PHY being resumed
when interface is not up"), here is the fix for SMSC911x (patch 2).
Also, I saw a splat running 'ifconfig' when interface was not up. Patch
1 fixes it. Thank to Geert for a comment on this one.

Patches are based on v6.2-rc3 and tested on a Renesas APE6-EK.

Changes since v1:
* patch 1 is not a revert but a proper fix this time


Wolfram Sang (2):
  smsc911x: only update stats when interface is up
  smsc911x: avoid PHY being resumed when interface is not up

 drivers/net/ethernet/smsc/smsc911x.c | 25 ++++++++++++++++---------
 1 file changed, 16 insertions(+), 9 deletions(-)

-- 
2.30.2

[PATCH net v2 1/2] smsc911x: only update stats when interface is up

[Wolfram Sang]

Otherwise the clocks are not enabled and reading registers will BUG.

Fixes: 1e30b8d755b8 ("net: smsc911x: Make Runtime PM handling more fine-grained")
Suggested-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
---
Changes since v1:
* patch is not a revert but a proper fix this time

 drivers/net/ethernet/smsc/smsc911x.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/smsc/smsc911x.c b/drivers/net/ethernet/smsc/smsc911x.c
index a2e511912e6a..67cb5eb9c716 100644
--- a/drivers/net/ethernet/smsc/smsc911x.c
+++ b/drivers/net/ethernet/smsc/smsc911x.c
@@ -1838,8 +1838,12 @@ smsc911x_hard_start_xmit(struct sk_buff *skb, struct net_device *dev)
 static struct net_device_stats *smsc911x_get_stats(struct net_device *dev)
 {
 	struct smsc911x_data *pdata = netdev_priv(dev);
-	smsc911x_tx_update_txcounters(dev);
-	dev->stats.rx_dropped += smsc911x_reg_read(pdata, RX_DROP);
+
+	if (netif_running(dev)) {
+		smsc911x_tx_update_txcounters(dev);
+		dev->stats.rx_dropped += smsc911x_reg_read(pdata, RX_DROP);
+	}
+
 	return &dev->stats;
 }
 
-- 
2.30.2

[PATCH net v2 2/2] smsc911x: avoid PHY being resumed when interface is not up

[Wolfram Sang]

SMSC911x doesn't need mdiobus suspend/resume, that's why it sets
'mac_managed_pm'. However, setting it needs to be moved from init to
probe, so mdiobus PM functions will really never be called (e.g. when
the interface is not up yet during suspend/resume). The errno is changed
because ENODEV has a special meaning when returned in probe().

Fixes: 3ce9f2bef755 ("net: smsc911x: Stop and start PHY during suspend and resume")
Suggested-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
---
Changes since v1:
* no change

In smsc911x_mii_probe(), I remove the sanity check for 'phydev' because
it was already done in smsc911x_mii_init(). Let me know if this is
acceptable or if a more defensive approach is favoured.


 drivers/net/ethernet/smsc/smsc911x.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/smsc/smsc911x.c b/drivers/net/ethernet/smsc/smsc911x.c
index 67cb5eb9c716..8b875bbbc05e 100644
--- a/drivers/net/ethernet/smsc/smsc911x.c
+++ b/drivers/net/ethernet/smsc/smsc911x.c
@@ -1019,12 +1019,7 @@ static int smsc911x_mii_probe(struct net_device *dev)
 	struct phy_device *phydev = NULL;
 	int ret;
 
-	/* find the first phy */
 	phydev = phy_find_first(pdata->mii_bus);
-	if (!phydev) {
-		netdev_err(dev, "no PHY found\n");
-		return -ENODEV;
-	}
 
 	SMSC_TRACE(pdata, probe, "PHY: addr %d, phy_id 0x%08X",
 		   phydev->mdio.addr, phydev->phy_id);
@@ -1037,8 +1032,6 @@ static int smsc911x_mii_probe(struct net_device *dev)
 		return ret;
 	}
 
-	/* Indicate that the MAC is responsible for managing PHY PM */
-	phydev->mac_managed_pm = true;
 	phy_attached_info(phydev);
 
 	phy_set_max_speed(phydev, SPEED_100);
@@ -1066,6 +1059,7 @@ static int smsc911x_mii_init(struct platform_device *pdev,
 			     struct net_device *dev)
 {
 	struct smsc911x_data *pdata = netdev_priv(dev);
+	struct phy_device *phydev;
 	int err = -ENXIO;
 
 	pdata->mii_bus = mdiobus_alloc();
@@ -1108,6 +1102,15 @@ static int smsc911x_mii_init(struct platform_device *pdev,
 		goto err_out_free_bus_2;
 	}
 
+	phydev = phy_find_first(pdata->mii_bus);
+	if (!phydev) {
+		netdev_err(dev, "no PHY found\n");
+		err = -ENOENT;
+		goto err_out_free_bus_2;
+	}
+
+	phydev->mac_managed_pm = true;
+
 	return 0;
 
 err_out_free_bus_2:
-- 
2.30.2

Re: [PATCH net v2 1/2] smsc911x: only update stats when interface is up

[Simon Horman]

On Mon, Mar 20, 2023 at 10:20:40AM +0100, Wolfram Sang wrote:
> Otherwise the clocks are not enabled and reading registers will BUG.
> 
> Fixes: 1e30b8d755b8 ("net: smsc911x: Make Runtime PM handling more fine-grained")
> Suggested-by: Geert Uytterhoeven <geert+renesas@glider.be>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

Reviewed-by: Simon Horman <simon.horman@corigine.com>

Re: [PATCH net v2 2/2] smsc911x: avoid PHY being resumed when interface is not up

[Simon Horman]

On Mon, Mar 20, 2023 at 10:20:41AM +0100, Wolfram Sang wrote:
> SMSC911x doesn't need mdiobus suspend/resume, that's why it sets
> 'mac_managed_pm'. However, setting it needs to be moved from init to
> probe, so mdiobus PM functions will really never be called (e.g. when
> the interface is not up yet during suspend/resume). The errno is changed
> because ENODEV has a special meaning when returned in probe().
> 
> Fixes: 3ce9f2bef755 ("net: smsc911x: Stop and start PHY during suspend and resume")
> Suggested-by: Heiner Kallweit <hkallweit1@gmail.com>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

Reviewed-by: Simon Horman <simon.horman@corigine.com>

Re: [PATCH net v2 2/2] smsc911x: avoid PHY being resumed when interface is not up

[Paolo Abeni]

On Mon, 2023-03-20 at 10:20 +0100, Wolfram Sang wrote:
> SMSC911x doesn't need mdiobus suspend/resume, that's why it sets
> 'mac_managed_pm'. However, setting it needs to be moved from init to
> probe, so mdiobus PM functions will really never be called (e.g. when
> the interface is not up yet during suspend/resume). The errno is changed
> because ENODEV has a special meaning when returned in probe().
> 
> Fixes: 3ce9f2bef755 ("net: smsc911x: Stop and start PHY during suspend and resume")
> Suggested-by: Heiner Kallweit <hkallweit1@gmail.com>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
> ---
> Changes since v1:
> * no change
> 
> In smsc911x_mii_probe(), I remove the sanity check for 'phydev' because
> it was already done in smsc911x_mii_init(). Let me know if this is
> acceptable or if a more defensive approach is favoured.

Since this is a fix, I would keep the old check, too.

> @@ -1108,6 +1102,15 @@ static int smsc911x_mii_init(struct platform_device *pdev,
>  		goto err_out_free_bus_2;
>  	}
>  
> +	phydev = phy_find_first(pdata->mii_bus);
> +	if (!phydev) {
> +		netdev_err(dev, "no PHY found\n");
> +		err = -ENOENT;
> +		goto err_out_free_bus_2;

Why don't you call mdiobus_unregister() in this error path?

mdiobus_register() completed successfully a few lines above.

Cheers,

Paolo

Re: [PATCH net v2 1/2] smsc911x: only update stats when interface is up

[Florian Fainelli]

On 3/20/23 02:20, Wolfram Sang wrote:
> Otherwise the clocks are not enabled and reading registers will BUG.
> 
> Fixes: 1e30b8d755b8 ("net: smsc911x: Make Runtime PM handling more fine-grained")
> Suggested-by: Geert Uytterhoeven <geert+renesas@glider.be>
> Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
-- 
Florian

Re: [PATCH net v2 2/2] smsc911x: avoid PHY being resumed when interface is not up

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 603 bytes --]

Hi Paolo,

> > In smsc911x_mii_probe(), I remove the sanity check for 'phydev' because
> > it was already done in smsc911x_mii_init(). Let me know if this is
> > acceptable or if a more defensive approach is favoured.
> 
> Since this is a fix, I would keep the old check, too.

Yes, makes sense.

> > +	phydev = phy_find_first(pdata->mii_bus);
> > +	if (!phydev) {
> > +		netdev_err(dev, "no PHY found\n");
> > +		err = -ENOENT;
> > +		goto err_out_free_bus_2;
> 
> Why don't you call mdiobus_unregister() in this error path?

Oversight. I will fix it.

Thank you for the review!


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]