Received-SPF: pass (google.com: domain of linux-kernel+bounces-15485-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Date: Wed, 3 Jan 2024 12:01:12 +0000
From: "Russell King (Oracle)" <linux@armlinux.org.uk>
To: Luiz Angelo Daros de Luca <luizluca@gmail.com>, kuba@kernel.org
Cc: netdev@vger.kernel.org, andrew@lunn.ch, hkallweit1@gmail.com,
	davem@davemloft.net, edumazet@google.com, pabeni@redhat.com,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH net-next] net: mdio: get/put device node during
 (un)registration
Message-ID: <ZZVMiPCoDyv/NZXN@shell.armlinux.org.uk>
References: <20231220045228.27079-2-luizluca@gmail.com>
 <ZZPtUIRerqTI2/yh@shell.armlinux.org.uk>
 <CAJq09z61JRNOBy6zLJ+D2pOVP-FCkofLjNghHSOkFJ=5q=6utQ@mail.gmail.com>
 <ZZU1SJlKpeU38c9I@shell.armlinux.org.uk>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZZU1SJlKpeU38c9I@shell.armlinux.org.uk>
Sender: Russell King (Oracle) <linux@armlinux.org.uk>

On Wed, Jan 03, 2024 at 10:22:00AM +0000, Russell King (Oracle) wrote:
> I agree with that approach, but as you rightly point out, we need MDIO
> to behave correctly, and I don't think that patching just one bit of
> MDIO to fix this mess is the right approach.

This is probably a safer approach to ensuring that the firmware data
reference count isn't dropped while the bus exists byensuring that we
always take a reference at register time. It also likely fixes similar
issues with ACPI and swnode based users as well.

It doesn't deal with the excess-refcount problem, as with this approach
the two issues are entirely independent of each other.

Please test to check that this addresses your issue. Thanks.

diff --git a/drivers/net/phy/mdio_bus.c b/drivers/net/phy/mdio_bus.c
index 6cf73c15635b..afbad1ad8683 100644
--- a/drivers/net/phy/mdio_bus.c
+++ b/drivers/net/phy/mdio_bus.c
@@ -193,6 +193,10 @@ static void mdiobus_release(struct device *d)
 	     bus->state != MDIOBUS_ALLOCATED,
 	     "%s: not in RELEASED or ALLOCATED state\n",
 	     bus->id);
+
+	if (bus->state == MDIOBUS_RELEASED)
+		fwnode_handle_put(dev_fwnode(d));
+
 	kfree(bus);
 }
 
@@ -684,6 +688,15 @@ int __mdiobus_register(struct mii_bus *bus, struct module *owner)
 	bus->dev.groups = NULL;
 	dev_set_name(&bus->dev, "%s", bus->id);
 
+	/* If the bus state is allocated, we're registering a fresh bus
+	 * that may have a fwnode associated with it. Grab a reference
+	 * to the fwnode. This will be dropped when the bus is released.
+	 * If the bus was set to unregistered, it means that the bus was
+	 * previously registered, and we've already grabbed a reference.
+	 */
+	if (bus->state == MDIOBUS_ALLOCATED)
+		fwnode_handle_get(dev_fwnode(&bus->dev));
+
 	/* We need to set state to MDIOBUS_UNREGISTERED to correctly release
 	 * the device in mdiobus_free()
 	 *
-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!