On Sat, May 16, 2015 at 02:32:33PM +0300, Dan Carpenter wrote:
> On Wed, May 13, 2015 at 01:22:34PM -0400, Benjamin Romer wrote:
> > @@ -1022,29 +997,21 @@ create_visor_device(struct visor_device *bdev,
> > goto away;
> > }
> >
> > - /* note: device_register is simply device_initialize + device_add */
> > - registered1 = true;
> > -
> > rc = register_devmajorminor_attributes(dev);
> > if (rc < 0) {
> > POSTCODE_LINUX_3(DEVICE_REGISTER_FAILURE_PC, chipset_dev_no,
> > DIAG_SEVERITY_ERR);
> > - goto away;
> > + goto away_register;
> > }
> >
> > - registered2 = true;
> > - rc = 0;
> > + list_add_tail(&dev->list_all, &list_all_device_instances);
> > + put_device(&dev->device);
>
> We don't want to call put_device() on the success path. It hopefully
> would have triggered an error if we call remove_visor_device()?
Yup, I agree. Mistake on my part when I was tweaking the failure paths
here. Thanks! Let me respin that patch.
Cheers,
Don
>
> > + return 0;
> >
> > +away_register:
> > + device_unregister(&dev->device);
> > away:
>
> regards,
> dan carpenter
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
