Hi,
On Mon, Apr 23, 2012 at 04:07:37PM +0200, Michael Grzeschik wrote:
> Call container_of after we have ensured that gadget is not NULL.
>
> Signed-off-by: Michael Grzeschik <m.grzeschik@xxxxxxxxxxxxxx>
> ---
> drivers/usb/gadget/ci13xxx_udc.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/usb/gadget/ci13xxx_udc.c b/drivers/usb/gadget/ci13xxx_udc.c
> index 9283ab0..7b03cf5 100644
> --- a/drivers/usb/gadget/ci13xxx_udc.c
> +++ b/drivers/usb/gadget/ci13xxx_udc.c
> @@ -1629,7 +1629,8 @@ __acquires(mEp->lock)
> static int _gadget_stop_activity(struct usb_gadget *gadget)
> {
> struct usb_ep *ep;
> - struct ci13xxx *udc = container_of(gadget, struct ci13xxx, gadget);
> + struct ci13xxx *udc;
> +
> unsigned long flags;
>
> trace("%p", gadget);
> @@ -1637,6 +1638,8 @@ static int _gadget_stop_activity(struct usb_gadget *gadget)
> if (gadget == NULL)
> return -EINVAL;
>
> + udc = container_of(gadget, struct ci13xxx, gadget);
There's no way for gadget to be NULL when we call
_gadget_stop_activity(), what you need is to drop the unnecessary gadget
== NULL check. If it happens to be NULL, it actually deserves to oops as
it would mean something is really, really broken (either the compiler or
the HW).
--
balbi
Attachment:
signature.asc
Description: Digital signature
