[PATCHv5 net 1/2] net: asix: fix modprobe "sysfs: cannot create duplicate filename"

[PATCHv5 net 1/2] net: asix: fix modprobe "sysfs: cannot create duplicate filename"

[Grant Grundler]

"modprobe asix ; rmmod asix ; modprobe asix" fails with:
   sysfs: cannot create duplicate filename \
   	'/devices/virtual/mdio_bus/usb-003:004'

Issue was originally reported by Anton Lundin on 2022-06-22 (link below).

Chrome OS team hit the same issue in Feb, 2023 when trying to find
work arounds for other issues with AX88172 devices.

The use of devm_mdiobus_register() with usbnet devices results in the
MDIO data being associated with the USB device. When the asix driver
is unloaded, the USB device continues to exist and the corresponding
"mdiobus_unregister()" is NOT called until the USB device is unplugged
or unauthorized. So the next "modprobe asix" will fail because the MDIO
phy sysfs attributes still exist.

The 'easy' (from a design PoV) fix is to use the non-devm variants of
mdiobus_* functions and explicitly manage this use in the asix_bind
and asix_unbind function calls. I've not explored trying to fix usbnet
initialization so devm_* stuff will work.

Fixes: e532a096be0e5 ("net: usb: asix: ax88772: add phylib support")
Reported-by: Anton Lundin <glance@acc.umu.se>
Link: https://lore.kernel.org/netdev/20220623063649.GD23685@pengutronix.de/T/
Tested-by: Eizan Miyamoto <eizan@chromium.org>
Signed-off-by: Grant Grundler <grundler@chromium.org>
---
 drivers/net/usb/asix_devices.c | 32 +++++++++++++++++++++++++++-----
 1 file changed, 27 insertions(+), 5 deletions(-)

V5: remove ax88772_mdio_unregister() call from error path in ax88772_init_phy()
    Jakub pointed out ax88772_mdio_unregister() will get called twice.
    (note: putting mdio handling in one function helps avoid this kind of bug.)

V4: add mdio_unregister to ax88172_bind() error handling paths

V3: rebase against netdev/net.git
    remove "TEST" prefix in subject line
    added Link: tag for Reported-by tag

V2: moved mdiobus_get_phy() call back into ax88772_init_phy()
   (Lukas Wunner is entirely correct this patch is much easier
   to backport without this patch hunk.)
   Added "Fixes:" tag per request from Florian Fainelli


diff --git a/drivers/net/usb/asix_devices.c b/drivers/net/usb/asix_devices.c
index 743cbf5d662c..f7cff58fe044 100644
--- a/drivers/net/usb/asix_devices.c
+++ b/drivers/net/usb/asix_devices.c
@@ -666,8 +666,9 @@ static int asix_resume(struct usb_interface *intf)
 static int ax88772_init_mdio(struct usbnet *dev)
 {
 	struct asix_common_private *priv = dev->driver_priv;
+	int ret;
 
-	priv->mdio = devm_mdiobus_alloc(&dev->udev->dev);
+	priv->mdio = mdiobus_alloc();
 	if (!priv->mdio)
 		return -ENOMEM;
 
@@ -679,7 +680,20 @@ static int ax88772_init_mdio(struct usbnet *dev)
 	snprintf(priv->mdio->id, MII_BUS_ID_SIZE, "usb-%03d:%03d",
 		 dev->udev->bus->busnum, dev->udev->devnum);
 
-	return devm_mdiobus_register(&dev->udev->dev, priv->mdio);
+	ret = mdiobus_register(priv->mdio);
+	if (ret) {
+		netdev_err(dev->net, "Could not register MDIO bus (err %d)\n", ret);
+		mdiobus_free(priv->mdio);
+		priv->mdio = NULL;
+	}
+
+	return ret;
+}
+
+static void ax88772_mdio_unregister(struct asix_common_private *priv)
+{
+	mdiobus_unregister(priv->mdio);
+	mdiobus_free(priv->mdio);
 }
 
 static int ax88772_init_phy(struct usbnet *dev)
@@ -896,16 +910,23 @@ static int ax88772_bind(struct usbnet *dev, struct usb_interface *intf)
 
 	ret = ax88772_init_mdio(dev);
 	if (ret)
-		return ret;
+		goto mdio_err;
 
 	ret = ax88772_phylink_setup(dev);
 	if (ret)
-		return ret;
+		goto phylink_err;
 
 	ret = ax88772_init_phy(dev);
 	if (ret)
-		phylink_destroy(priv->phylink);
+		goto initphy_err;
 
+	return 0;
+
+initphy_err:
+	phylink_destroy(priv->phylink);
+phylink_err:
+	ax88772_mdio_unregister(priv);
+mdio_err:
 	return ret;
 }
 
@@ -926,6 +947,7 @@ static void ax88772_unbind(struct usbnet *dev, struct usb_interface *intf)
 	phylink_disconnect_phy(priv->phylink);
 	rtnl_unlock();
 	phylink_destroy(priv->phylink);
+	ax88772_mdio_unregister(priv);
 	asix_rx_fixup_common_free(dev->driver_priv);
 }
 
-- 
2.40.0.rc1.284.g88254d51c5-goog

Re: [PATCHv5 net 1/2] net: asix: fix modprobe "sysfs: cannot create duplicate filename"

[Simon Horman]

On Fri, Mar 17, 2023 at 11:13:21AM -0700, Grant Grundler wrote:
> "modprobe asix ; rmmod asix ; modprobe asix" fails with:
>    sysfs: cannot create duplicate filename \
>    	'/devices/virtual/mdio_bus/usb-003:004'
> 
> Issue was originally reported by Anton Lundin on 2022-06-22 (link below).
> 
> Chrome OS team hit the same issue in Feb, 2023 when trying to find
> work arounds for other issues with AX88172 devices.
> 
> The use of devm_mdiobus_register() with usbnet devices results in the
> MDIO data being associated with the USB device. When the asix driver
> is unloaded, the USB device continues to exist and the corresponding
> "mdiobus_unregister()" is NOT called until the USB device is unplugged
> or unauthorized. So the next "modprobe asix" will fail because the MDIO
> phy sysfs attributes still exist.
> 
> The 'easy' (from a design PoV) fix is to use the non-devm variants of
> mdiobus_* functions and explicitly manage this use in the asix_bind
> and asix_unbind function calls. I've not explored trying to fix usbnet
> initialization so devm_* stuff will work.
> 
> Fixes: e532a096be0e5 ("net: usb: asix: ax88772: add phylib support")
> Reported-by: Anton Lundin <glance@acc.umu.se>
> Link: https://lore.kernel.org/netdev/20220623063649.GD23685@pengutronix.de/T/
> Tested-by: Eizan Miyamoto <eizan@chromium.org>
> Signed-off-by: Grant Grundler <grundler@chromium.org>

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

Re: [PATCHv5 net 1/2] net: asix: fix modprobe "sysfs: cannot create duplicate filename"

[Jakub Kicinski]

On Fri, 17 Mar 2023 11:13:21 -0700 Grant Grundler wrote:
> "modprobe asix ; rmmod asix ; modprobe asix" fails with:
>    sysfs: cannot create duplicate filename \
>    	'/devices/virtual/mdio_bus/usb-003:004'

Code looks good but now you posted only one patch with [PATCH 1/2]
so patchwork is dutifully waiting to receive [PATCH 2/2]. And not
running any checks. One more time, please :) You can keep Simon's
review tag.