net: dsa: be compatible with masters which unregister on shutdown

Lino reports that on his system with bcmgenet as DSA master and KSZ9897
as a switch, rebooting or shutting down never works properly.

What does the bcmgenet driver have special to trigger this, that other
DSA masters do not? It has an implementation of ->shutdown which simply
calls its ->remove implementation. Otherwise said, it unregisters its
network interface on shutdown.

This message can be seen in a loop, and it hangs the reboot process there:

unregister_netdevice: waiting for eth0 to become free. Usage count = 3

So why 3?

A usage count of 1 is normal for a registered network interface, and any
virtual interface which links itself as an upper of that will increment
it via dev_hold. In the case of DSA, this is the call path:

dsa_slave_create
-> netdev_upper_dev_link
   -> __netdev_upper_dev_link
      -> __netdev_adjacent_dev_insert
         -> dev_hold

So a DSA switch with 3 interfaces will result in a usage count elevated
by two, and netdev_wait_allrefs will wait until they have gone away.

Other stacked interfaces, like VLAN, watch NETDEV_UNREGISTER events and
delete themselves, but DSA cannot just vanish and go poof, at most it
can unbind itself from the switch devices, but that must happen strictly
earlier compared to when the DSA master unregisters its net_device, so
reacting on the NETDEV_UNREGISTER event is way too late.

It seems that it is a pretty established pattern to have a driver's
->shutdown hook redirect to its ->remove hook, so the same code is
executed regardless of whether the driver is unbound from the device, or
the system is just shutting down. As Florian puts it, it is quite a big
hammer for bcmgenet to unregister its net_device during shutdown, but
having a common code path with the driver unbind helps ensure it is well
tested.

So DSA, for better or for worse, has to live with that and engage in an
arms race of implementing the ->shutdown hook too, from all individual
drivers, and do something sane when paired with masters that unregister
their net_device there. The only sane thing to do, of course, is to
unlink from the master.

However, complications arise really quickly.

The pattern of redirecting ->shutdown to ->remove is not unique to
bcmgenet or even to net_device drivers. In fact, SPI controllers do it
too (see dspi_shutdown -> dspi_remove), and presumably, I2C controllers
and MDIO controllers do it too (this is something I have not researched
too deeply, but even if this is not the case today, it is certainly
plausible to happen in the future, and must be taken into consideration).

Since DSA switches might be SPI devices, I2C devices, MDIO devices, the
insane implication is that for the exact same DSA switch device, we
might have both ->shutdown and ->remove getting called.

So we need to do something with that insane environment. The pattern
I've come up with is "if this, then not that", so if either ->shutdown
or ->remove gets called, we set the device's drvdata to NULL, and in the
other hook, we check whether the drvdata is NULL and just do nothing.
This is probably not necessary for platform devices, just for devices on
buses, but I would really insist for consistency among drivers, because
when code is copy-pasted, it is not always copy-pasted from the best
sources.

So depending on whether the DSA switch's ->remove or ->shutdown will get
called first, we cannot really guarantee even for the same driver if
rebooting will result in the same code path on all platforms. But
nonetheless, we need to do something minimally reasonable on ->shutdown
too to fix the bug. Of course, the ->remove will do more (a full
teardown of the tree, with all data structures freed, and this is why
the bug was not caught for so long). The new ->shutdown method is kept
separate from dsa_unregister_switch not because we couldn't have
unregistered the switch, but simply in the interest of doing something
quick and to the point.

The big question is: does the DSA switch's ->shutdown get called earlier
than the DSA master's ->shutdown? If not, there is still a risk that we
might still trigger the WARN_ON in unregister_netdevice that says we are
attempting to unregister a net_device which has uppers. That's no good.
Although the reference to the master net_device won't physically go away
even if DSA's ->shutdown comes afterwards, remember we have a dev_hold
on it.

The answer to that question lies in this comment above device_link_add:

 * A side effect of the link creation is re-ordering of dpm_list and the
 * devices_kset list by moving the consumer device and all devices depending
 * on it to the ends of these lists (that does not happen to devices that have
 * not been registered when this function is called).

so the fact that DSA uses device_link_add towards its master is not
exactly for nothing. device_shutdown() walks devices_kset from the back,
so this is our guarantee that DSA's shutdown happens before the master's
shutdown.

Fixes: 2f1e8ea ("net: dsa: link interfaces with the DSA master to get rid of lockdep warnings")
Link: https://lore.kernel.org/netdev/[email protected]/
Reported-by: Lino Sanfilippo <[email protected]>
Signed-off-by: Vladimir Oltean <[email protected]>
Tested-by: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Author: vladimiroltean

drivers/net/dsa/b53/b53_mdio.c

@@ -351,9 +351,25 @@ static int b53_mdio_probe(struct mdio_device *mdiodev)
 static void b53_mdio_remove(struct mdio_device *mdiodev)
 {
 	struct b53_device *dev = dev_get_drvdata(&mdiodev->dev);
-	struct dsa_switch *ds = dev->ds;
 
-	dsa_unregister_switch(ds);
+	if (!dev)
+		return;
+
+	b53_switch_remove(dev);
+
+	dev_set_drvdata(&mdiodev->dev, NULL);
+}
+
+static void b53_mdio_shutdown(struct mdio_device *mdiodev)
+{
+	struct b53_device *dev = dev_get_drvdata(&mdiodev->dev);
+
+	if (!dev)
+		return;
+
+	b53_switch_shutdown(dev);
+
+	dev_set_drvdata(&mdiodev->dev, NULL);
 }
 
 static const struct of_device_id b53_of_match[] = {
@@ -373,6 +389,7 @@ MODULE_DEVICE_TABLE(of, b53_of_match);
 static struct mdio_driver b53_mdio_driver = {
 	.probe	= b53_mdio_probe,
 	.remove	= b53_mdio_remove,
+	.shutdown = b53_mdio_shutdown,
 	.mdiodrv.driver = {
 		.name = "bcm53xx",
 		.of_match_table = b53_of_match,

drivers/net/dsa/b53/b53_mmap.c

@@ -316,9 +316,21 @@ static int b53_mmap_remove(struct platform_device *pdev)
 	if (dev)
 		b53_switch_remove(dev);
 
+	platform_set_drvdata(pdev, NULL);
+
 	return 0;
 }
 
+static void b53_mmap_shutdown(struct platform_device *pdev)
+{
+	struct b53_device *dev = platform_get_drvdata(pdev);
+
+	if (dev)
+		b53_switch_shutdown(dev);
+
+	platform_set_drvdata(pdev, NULL);
+}
+
 static const struct of_device_id b53_mmap_of_table[] = {
 	{ .compatible = "brcm,bcm3384-switch" },
 	{ .compatible = "brcm,bcm6328-switch" },
@@ -331,6 +343,7 @@ MODULE_DEVICE_TABLE(of, b53_mmap_of_table);
 static struct platform_driver b53_mmap_driver = {
 	.probe = b53_mmap_probe,
 	.remove = b53_mmap_remove,
+	.shutdown = b53_mmap_shutdown,
 	.driver = {
 		.name = "b53-switch",
 		.of_match_table = b53_mmap_of_table,

drivers/net/dsa/b53/b53_priv.h

@@ -228,6 +228,11 @@ static inline void b53_switch_remove(struct b53_device *dev)
 	dsa_unregister_switch(dev->ds);
 }
 
+static inline void b53_switch_shutdown(struct b53_device *dev)
+{
+	dsa_switch_shutdown(dev->ds);
+}
+
 #define b53_build_op(type_op_size, val_type)				\
 static inline int b53_##type_op_size(struct b53_device *dev, u8 page,	\
 				     u8 reg, val_type val)		\

drivers/net/dsa/b53/b53_spi.c

@@ -321,9 +321,21 @@ static int b53_spi_remove(struct spi_device *spi)
 	if (dev)
 		b53_switch_remove(dev);
 
+	spi_set_drvdata(spi, NULL);
+
 	return 0;
 }
 
+static void b53_spi_shutdown(struct spi_device *spi)
+{
+	struct b53_device *dev = spi_get_drvdata(spi);
+
+	if (dev)
+		b53_switch_shutdown(dev);
+
+	spi_set_drvdata(spi, NULL);
+}
+
 static const struct of_device_id b53_spi_of_match[] = {
 	{ .compatible = "brcm,bcm5325" },
 	{ .compatible = "brcm,bcm5365" },
@@ -344,6 +356,7 @@ static struct spi_driver b53_spi_driver = {
 	},
 	.probe	= b53_spi_probe,
 	.remove	= b53_spi_remove,
+	.shutdown = b53_spi_shutdown,
 };
 
 module_spi_driver(b53_spi_driver);

drivers/net/dsa/b53/b53_srab.c

@@ -629,17 +629,34 @@ static int b53_srab_probe(struct platform_device *pdev)
 static int b53_srab_remove(struct platform_device *pdev)
 {
 	struct b53_device *dev = platform_get_drvdata(pdev);
-	struct b53_srab_priv *priv = dev->priv;
 
-	b53_srab_intr_set(priv, false);
+	if (!dev)
+		return 0;
+
+	b53_srab_intr_set(dev->priv, false);
 	b53_switch_remove(dev);
 
+	platform_set_drvdata(pdev, NULL);
+
 	return 0;
 }
 
+static void b53_srab_shutdown(struct platform_device *pdev)
+{
+	struct b53_device *dev = platform_get_drvdata(pdev);
+
+	if (!dev)
+		return;
+
+	b53_switch_shutdown(dev);
+
+	platform_set_drvdata(pdev, NULL);
+}
+
 static struct platform_driver b53_srab_driver = {
 	.probe = b53_srab_probe,
 	.remove = b53_srab_remove,
+	.shutdown = b53_srab_shutdown,
 	.driver = {
 		.name = "b53-srab-switch",
 		.of_match_table = b53_srab_of_match,

drivers/net/dsa/bcm_sf2.c

@@ -1512,6 +1512,9 @@ static int bcm_sf2_sw_remove(struct platform_device *pdev)
 {
 	struct bcm_sf2_priv *priv = platform_get_drvdata(pdev);
 
+	if (!priv)
+		return 0;
+
 	priv->wol_ports_mask = 0;
 	/* Disable interrupts */
 	bcm_sf2_intr_disable(priv);
@@ -1523,13 +1526,18 @@ static int bcm_sf2_sw_remove(struct platform_device *pdev)
 	if (priv->type == BCM7278_DEVICE_ID)
 		reset_control_assert(priv->rcdev);
 
+	platform_set_drvdata(pdev, NULL);
+
 	return 0;
 }
 
 static void bcm_sf2_sw_shutdown(struct platform_device *pdev)
 {
 	struct bcm_sf2_priv *priv = platform_get_drvdata(pdev);
 
+	if (!priv)
+		return;
+
 	/* For a kernel about to be kexec'd we want to keep the GPHY on for a
 	 * successful MDIO bus scan to occur. If we did turn off the GPHY
 	 * before (e.g: port_disable), this will also power it back on.
@@ -1538,6 +1546,10 @@ static void bcm_sf2_sw_shutdown(struct platform_device *pdev)
 	 */
 	if (priv->hw_params.num_gphy == 1)
 		bcm_sf2_gphy_enable_set(priv->dev->ds, true);
+
+	dsa_switch_shutdown(priv->dev->ds);
+
+	platform_set_drvdata(pdev, NULL);
 }
 
 #ifdef CONFIG_PM_SLEEP

drivers/net/dsa/dsa_loop.c

@@ -340,10 +340,29 @@ static int dsa_loop_drv_probe(struct mdio_device *mdiodev)
 static void dsa_loop_drv_remove(struct mdio_device *mdiodev)
 {
 	struct dsa_switch *ds = dev_get_drvdata(&mdiodev->dev);
-	struct dsa_loop_priv *ps = ds->priv;
+	struct dsa_loop_priv *ps;
+
+	if (!ds)
+		return;
+
+	ps = ds->priv;
 
 	dsa_unregister_switch(ds);
 	dev_put(ps->netdev);
+
+	dev_set_drvdata(&mdiodev->dev, NULL);
+}
+
+static void dsa_loop_drv_shutdown(struct mdio_device *mdiodev)
+{
+	struct dsa_switch *ds = dev_get_drvdata(&mdiodev->dev);
+
+	if (!ds)
+		return;
+
+	dsa_switch_shutdown(ds);
+
+	dev_set_drvdata(&mdiodev->dev, NULL);
 }
 
 static struct mdio_driver dsa_loop_drv = {
@@ -352,6 +371,7 @@ static struct mdio_driver dsa_loop_drv = {
 	},
 	.probe	= dsa_loop_drv_probe,
 	.remove	= dsa_loop_drv_remove,
+	.shutdown = dsa_loop_drv_shutdown,
 };
 
 #define NUM_FIXED_PHYS	(DSA_LOOP_NUM_PORTS - 2)

drivers/net/dsa/lan9303-core.c

@@ -1379,6 +1379,12 @@ int lan9303_remove(struct lan9303 *chip)
 }
 EXPORT_SYMBOL(lan9303_remove);
 
+void lan9303_shutdown(struct lan9303 *chip)
+{
+	dsa_switch_shutdown(chip->ds);
+}
+EXPORT_SYMBOL(lan9303_shutdown);
+
 MODULE_AUTHOR("Juergen Borleis <[email protected]>");
 MODULE_DESCRIPTION("Core driver for SMSC/Microchip LAN9303 three port ethernet switch");
 MODULE_LICENSE("GPL v2");

drivers/net/dsa/lan9303.h

@@ -10,3 +10,4 @@ extern const struct lan9303_phy_ops lan9303_indirect_phy_ops;
 
 int lan9303_probe(struct lan9303 *chip, struct device_node *np);
 int lan9303_remove(struct lan9303 *chip);
+void lan9303_shutdown(struct lan9303 *chip);

drivers/net/dsa/lan9303_i2c.c

@@ -67,13 +67,28 @@ static int lan9303_i2c_probe(struct i2c_client *client,
 
 static int lan9303_i2c_remove(struct i2c_client *client)
 {
-	struct lan9303_i2c *sw_dev;
+	struct lan9303_i2c *sw_dev = i2c_get_clientdata(client);
 
-	sw_dev = i2c_get_clientdata(client);
 	if (!sw_dev)
-		return -ENODEV;
+		return 0;
+
+	lan9303_remove(&sw_dev->chip);
+
+	i2c_set_clientdata(client, NULL);
+
+	return 0;
+}
+
+static void lan9303_i2c_shutdown(struct i2c_client *client)
+{
+	struct lan9303_i2c *sw_dev = i2c_get_clientdata(client);
+
+	if (!sw_dev)
+		return;
+
+	lan9303_shutdown(&sw_dev->chip);
 
-	return lan9303_remove(&sw_dev->chip);
+	i2c_set_clientdata(client, NULL);
 }
 
 /*-------------------------------------------------------------------------*/
@@ -97,6 +112,7 @@ static struct i2c_driver lan9303_i2c_driver = {
 	},
 	.probe = lan9303_i2c_probe,
 	.remove = lan9303_i2c_remove,
+	.shutdown = lan9303_i2c_shutdown,
 	.id_table = lan9303_i2c_id,
 };
 module_i2c_driver(lan9303_i2c_driver);

drivers/net/dsa/lan9303_mdio.c

@@ -138,6 +138,20 @@ static void lan9303_mdio_remove(struct mdio_device *mdiodev)
 		return;
 
 	lan9303_remove(&sw_dev->chip);
+
+	dev_set_drvdata(&mdiodev->dev, NULL);
+}
+
+static void lan9303_mdio_shutdown(struct mdio_device *mdiodev)
+{
+	struct lan9303_mdio *sw_dev = dev_get_drvdata(&mdiodev->dev);
+
+	if (!sw_dev)
+		return;
+
+	lan9303_shutdown(&sw_dev->chip);
+
+	dev_set_drvdata(&mdiodev->dev, NULL);
 }
 
 /*-------------------------------------------------------------------------*/
@@ -155,6 +169,7 @@ static struct mdio_driver lan9303_mdio_driver = {
 	},
 	.probe  = lan9303_mdio_probe,
 	.remove = lan9303_mdio_remove,
+	.shutdown = lan9303_mdio_shutdown,
 };
 mdio_module_driver(lan9303_mdio_driver);
 

drivers/net/dsa/lantiq_gswip.c

@@ -2184,6 +2184,9 @@ static int gswip_remove(struct platform_device *pdev)
 	struct gswip_priv *priv = platform_get_drvdata(pdev);
 	int i;
 
+	if (!priv)
+		return 0;
+
 	/* disable the switch */
 	gswip_mdio_mask(priv, GSWIP_MDIO_GLOB_ENABLE, 0, GSWIP_MDIO_GLOB);
 
@@ -2197,9 +2200,23 @@ static int gswip_remove(struct platform_device *pdev)
 	for (i = 0; i < priv->num_gphy_fw; i++)
 		gswip_gphy_fw_remove(priv, &priv->gphy_fw[i]);
 
+	platform_set_drvdata(pdev, NULL);
+
 	return 0;
 }
 
+static void gswip_shutdown(struct platform_device *pdev)
+{
+	struct gswip_priv *priv = platform_get_drvdata(pdev);
+
+	if (!priv)
+		return;
+
+	dsa_switch_shutdown(priv->ds);
+
+	platform_set_drvdata(pdev, NULL);
+}
+
 static const struct gswip_hw_info gswip_xrx200 = {
 	.max_ports = 7,
 	.cpu_port = 6,
@@ -2223,6 +2240,7 @@ MODULE_DEVICE_TABLE(of, gswip_of_match);
 static struct platform_driver gswip_driver = {
 	.probe = gswip_probe,
 	.remove = gswip_remove,
+	.shutdown = gswip_shutdown,
 	.driver = {
 		.name = "gswip",
 		.of_match_table = gswip_of_match,

drivers/net/dsa/microchip/ksz8795_spi.c

@@ -94,15 +94,24 @@ static int ksz8795_spi_remove(struct spi_device *spi)
 	if (dev)
 		ksz_switch_remove(dev);
 
+	spi_set_drvdata(spi, NULL);
+
 	return 0;
 }
 
 static void ksz8795_spi_shutdown(struct spi_device *spi)
 {
 	struct ksz_device *dev = spi_get_drvdata(spi);
 
-	if (dev && dev->dev_ops->shutdown)
+	if (!dev)
+		return;
+
+	if (dev->dev_ops->shutdown)
 		dev->dev_ops->shutdown(dev);
+
+	dsa_switch_shutdown(dev->ds);
+
+	spi_set_drvdata(spi, NULL);
 }
 
 static const struct of_device_id ksz8795_dt_ids[] = {