On Wed, Jan 29, 2020 at 08:30:37PM +0300, Dan Carpenter wrote:
> The "actual_length" variable might be uninitialized on some failure
> paths. It's harmless but static analysis tools like Smatch complain
> and at runtime the UBSan tool will likely complain as well.
>
> Fixes: e7542bc382f8 ("USB: serial: ir-usb: make set_termios synchronous")
> Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
Nice catch, I'll queue this up for 5.6-rc.
> ---
> drivers/usb/serial/ir-usb.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/usb/serial/ir-usb.c b/drivers/usb/serial/ir-usb.c
> index 79d0586e2b33..172261a908d8 100644
> --- a/drivers/usb/serial/ir-usb.c
> +++ b/drivers/usb/serial/ir-usb.c
> @@ -448,7 +448,7 @@ static void ir_set_termios(struct tty_struct *tty,
> usb_sndbulkpipe(udev, port->bulk_out_endpointAddress),
> transfer_buffer, 1, &actual_length, 5000);
> if (ret || actual_length != 1) {
> - if (actual_length != 1)
> + if (!ret)
> ret = -EIO;
> dev_err(&port->dev, "failed to change line speed: %d\n", ret);
> }
Johan
