On Tue, Jul 24, 2018 at 08:09:59PM +0200, Marcus Folkesson wrote:

Hello Dmitry,

On Tue, Jul 24, 2018 at 02:38:04AM +0000, Dmitry Torokhov wrote:

quoted

Hi Marcus,

On Mon, Jul 16, 2018 at 04:40:14PM +0200, Marcus Folkesson wrote:

quoted

The USB device is only needed during setup, so put it back after
initialization and do not store it in our private struct.

Also, the USB device is a parent of USB interface so our driver
model rules ensure that USB device should not disappear while
interface device is still there.
So not keep a refcount on the device is safe.

Reported-by: Alexey Khoroshilov <redacted>
Signed-off-by: Marcus Folkesson <marcus.folkesson@gmail.com>
---
 drivers/input/joystick/pxrc.c | 22 ++++++++++++----------
 1 file changed, 12 insertions(+), 10 deletions(-)

diff --git a/drivers/input/joystick/pxrc.c b/drivers/input/joystick/pxrc.c
index 07a0dbd3ced2..46a7acb747bf 100644
--- a/drivers/input/joystick/pxrc.c
+++ b/drivers/input/joystick/pxrc.c

...

quoted

@@ -204,23 +204,25 @@ static int pxrc_probe(struct usb_interface *intf,
 		return -ENOMEM;
 
 	mutex_init(&pxrc->pm_mutex);
-	pxrc->udev = usb_get_dev(interface_to_usbdev(intf));
+	udev = usb_get_dev(interface_to_usbdev(intf));

There is really no need to "get" device for the probe duration, or in
general, when you are not storing the reference to it.

I posted series with an updated version of this patch plus couple more
cleanups/fixes, and would appreciate if you could give it a spin.

Thank you for doing this.

I have reviewed the patchset and tested on real hardware, and it looks good
to me.

For what it's worth:

Reviewed-by: Marcus Folkesson <marcus.folkesson@gmail.com>
Tested-by: Marcus Folkesson <marcus.folkesson@gmail.com> 

On the whole patchset.

Excellent, thank you.

-- 
Dmitry