On 01/03/20 13:58:08, Dan Carpenter wrote:
> On Fri, Dec 20, 2019 at 09:14:59PM +0000, Malcolm Priestley wrote:
> > Starting with commit 59608cb1de1856
> > ("staging: vt6656: clean function's error path in usbpipe.c")
> > the usb control functions have returned errors throughout driver
> > with only logical variable checking.
>
> Use the Fixes tag for this.
>
> Fixes: 59608cb1de18 ("staging: vt6656: clean function's error path in usbpipe.c")
>
> 12 digits to the hash. Add Quentin to the CC list.
>
> >
> > However, usb_control_msg return the amount of bytes transferred
> > this means that normal operation causes errors.
> >
> > Correct the return function so only return zero when transfer
> > is successful.
> >
> > Cc: stable <stable@xxxxxxxxxxxxxxx> # v5.3+
> > Signed-off-by: Malcolm Priestley <tvboxspy@xxxxxxxxx>
> > ---
> > drivers/staging/vt6656/usbpipe.c | 8 ++++++--
> > 1 file changed, 6 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/staging/vt6656/usbpipe.c b/drivers/staging/vt6656/usbpipe.c
> > index d3304df6bd53..488ebd98773d 100644
> > --- a/drivers/staging/vt6656/usbpipe.c
> > +++ b/drivers/staging/vt6656/usbpipe.c
> > @@ -59,7 +59,9 @@ int vnt_control_out(struct vnt_private *priv, u8 request, u16 value,
> >
> > kfree(usb_buffer);
> >
> > - if (ret >= 0 && ret < (int)length)
> > + if (ret == (int)length)
>
> No need for this cast (no need in the original either).
>
> > + ret = 0;
> > + else
> > ret = -EIO;
>
> It would be better to preserve the error codes from usb_control_msg().
>
> if (ret == length)
> ret = 0;
> else if (ret >= 0)
> ret = -EIO;
>
> regards,
> dan carpenter
>
Thanks for CC.
Nice catch. Dan is right, we should forward any error code from
usb_control_msg().
Regards,
Quentin Deslandes
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
