| Message ID | E1qNxvu-00111m-1V@rmk-PC.armlinux.org.uk (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 09bd2d7ddaedcdfa4aa3f4d54c103f984a93bd32 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net-next] net: mdio_bus: validate "addr" for mdiobus_is_registered_device() | expand |
On Mon, Jul 24, 2023 at 04:57:14PM +0100, Russell King (Oracle) wrote: > mdiobus_is_registered_device() doesn't checking that "addr" was valid > before dereferencing bus->mdio_map[]. Extract the code that checks > this from mdiobus_get_phy(), and use it here as well. > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Andrew
Hello: This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <kuba@kernel.org>: On Mon, 24 Jul 2023 16:57:14 +0100 you wrote: > mdiobus_is_registered_device() doesn't checking that "addr" was valid > before dereferencing bus->mdio_map[]. Extract the code that checks > this from mdiobus_get_phy(), and use it here as well. > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> > --- > I've had this patch kicking about for a while - probably a good idea? > It shouldn't cause a regression, but if it does it means we're already > dereferencing the arrray outside its bounds in > mdiobus_is_registered_device(). > > [...] Here is the summary with links: - [net-next] net: mdio_bus: validate "addr" for mdiobus_is_registered_device() https://git.kernel.org/netdev/net-next/c/09bd2d7ddaed You are awesome, thank you!
diff --git a/drivers/net/phy/mdio_bus.c b/drivers/net/phy/mdio_bus.c index bc04048de2fa..25dcaa49ab8b 100644 --- a/drivers/net/phy/mdio_bus.c +++ b/drivers/net/phy/mdio_bus.c @@ -107,16 +107,21 @@ int mdiobus_unregister_device(struct mdio_device *mdiodev) } EXPORT_SYMBOL(mdiobus_unregister_device); -struct phy_device *mdiobus_get_phy(struct mii_bus *bus, int addr) +static struct mdio_device *mdiobus_find_device(struct mii_bus *bus, int addr) { bool addr_valid = addr >= 0 && addr < ARRAY_SIZE(bus->mdio_map); - struct mdio_device *mdiodev; if (WARN_ONCE(!addr_valid, "addr %d out of range\n", addr)) return NULL; - mdiodev = bus->mdio_map[addr]; + return bus->mdio_map[addr]; +} +struct phy_device *mdiobus_get_phy(struct mii_bus *bus, int addr) +{ + struct mdio_device *mdiodev; + + mdiodev = mdiobus_find_device(bus, addr); if (!mdiodev) return NULL; @@ -129,7 +134,7 @@ EXPORT_SYMBOL(mdiobus_get_phy); bool mdiobus_is_registered_device(struct mii_bus *bus, int addr) { - return bus->mdio_map[addr]; + return mdiobus_find_device(bus, addr) != NULL; } EXPORT_SYMBOL(mdiobus_is_registered_device);
mdiobus_is_registered_device() doesn't checking that "addr" was valid before dereferencing bus->mdio_map[]. Extract the code that checks this from mdiobus_get_phy(), and use it here as well. Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> --- I've had this patch kicking about for a while - probably a good idea? It shouldn't cause a regression, but if it does it means we're already dereferencing the arrray outside its bounds in mdiobus_is_registered_device(). drivers/net/phy/mdio_bus.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)