Re: [PATCH] phy: tegra: xusb: Fix use-after-free issue

From: Vinod Koul
Date: Mon May 08 2023 - 04:57:13 EST


On 04-05-23, 11:15, Thierry Reding wrote:
> On Thu, May 04, 2023 at 01:54:46PM +0800, Haotien Hsu wrote:
> > From: EJ Hsu <ejh@xxxxxxxxxx>
> >
> > For the dual-role port, it will assign the phy dev to usb-phy dev and
> > use the port dev driver as the dev driver of usb-phy.
> >
> > When we try to destroy the port dev, it will destroy its dev driver
> > as well. But we did not remove the reference from usb-phy dev. This
> > might cause the use-after-free issue in KASAN.
> >
> > Fixes: e8f7d2f409a1 ("phy: tegra: xusb: Add usb-phy support")
> > Cc: stable@xxxxxxxxxxxxxxx
> >
> > Signed-off-by: EJ Hsu <ejh@xxxxxxxxxx>
> > Signed-off-by: Haotien Hsu <haotienh@xxxxxxxxxx>
> > ---
> > drivers/phy/tegra/xusb.c | 15 ++++++++-------
> > 1 file changed, 8 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/phy/tegra/xusb.c b/drivers/phy/tegra/xusb.c
> > index 78045bd6c214..515298a9a433 100644
> > --- a/drivers/phy/tegra/xusb.c
> > +++ b/drivers/phy/tegra/xusb.c
> > @@ -563,13 +563,6 @@ static int tegra_xusb_port_init(struct tegra_xusb_port *port,
> >
> > static void tegra_xusb_port_unregister(struct tegra_xusb_port *port)
> > {
> > - if (!IS_ERR_OR_NULL(port->usb_role_sw)) {
> > - of_platform_depopulate(&port->dev);
> > - usb_role_switch_unregister(port->usb_role_sw);
> > - cancel_work_sync(&port->usb_phy_work);
> > - usb_remove_phy(&port->usb_phy);
> > - }
> > -
> > if (port->ops->remove)
> > port->ops->remove(port);
> >
> > @@ -832,6 +825,14 @@ void tegra_xusb_usb2_port_remove(struct tegra_xusb_port *port)
> > {
> > struct tegra_xusb_usb2_port *usb2 = to_usb2_port(port);
> >
> > + if (!IS_ERR_OR_NULL(port->usb_role_sw)) {
> > + of_platform_depopulate(&port->dev);
> > + usb_role_switch_unregister(port->usb_role_sw);
> > + cancel_work_sync(&port->usb_phy_work);
> > + usb_remove_phy(&port->usb_phy);
> > + port->usb_phy.dev->driver = NULL;
> > + }
> > +
>
> It's a bit difficult to tell what exactly changed because of the
> additional move. The move looks correct because this is only valid for
> USB2 ports, but I think it'd be easier to review if the move was done
> prior to the addition of the ->driver clearing. Or perhaps a mention in
> the commit message would be nice.
>
> It's obviously up to Vinod and Kishon if they want to see this
> clarified. Either way is fine with me, so:

I agree changes need to be clarified, expecting updated v2 :-)

>
> Acked-by: Thierry Reding <treding@xxxxxxxxxx>



--
~Vinod