staging: kpc2000: added missing clean-up to probe_core_uio.

On error, probe_core_uio just returned an error without freeing
resources which had previously been allocated.  Added the missing
clean-up code.

Updated TODO.
Signed-off-by: Jeremy Sowden <jeremy@azazel.net>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/staging/kpc2000/TODO

 - the kpc_spi driver doesn't seem to let multiple transactions (to different instances of the core) happen in parallel...
 - The kpc_i2c driver is a hot mess, it should probably be cleaned up a ton.  It functions against current hardware though.
 - pcard->card_num in kp2000_pcie_probe() is a global variable and needs atomic / locking / something better.
-- probe_core_uio() probably needs error handling
 - the loop in kp2000_probe_cores() that uses probe_core_uio() also probably needs error handling
 - would be nice if the AIO fileops in kpc_dma could be made to work
     - probably want to add a CONFIG_ option to control compilation of the AIO functions

drivers/staging/kpc2000/kpc2000/cell_probe.c

@@ -295,6 +295,7 @@ static int probe_core_uio(unsigned int core_num, struct kp2000_device *pcard,
     kudev->dev = device_create(kpc_uio_class, &pcard->pdev->dev, MKDEV(0,0), kudev, "%s.%d.%d.%d", kudev->uioinfo.name, pcard->card_num, cte.type, kudev->core_num);
     if (IS_ERR(kudev->dev)) {
         dev_err(&pcard->pdev->dev, "probe_core_uio device_create failed!\n");
+        kfree(kudev);
         return -ENODEV;
     }
     dev_set_drvdata(kudev->dev, kudev);
@@ -302,6 +303,8 @@ static int probe_core_uio(unsigned int core_num, struct kp2000_device *pcard,
     rv = uio_register_device(kudev->dev, &kudev->uioinfo);
     if (rv){
         dev_err(&pcard->pdev->dev, "probe_core_uio failed uio_register_device: %d\n", rv);
+        put_device(kudev->dev);
+        kfree(kudev);
         return rv;
     }