Re: USB 2.0: No giveback comes for one submitted URB

[Alan Stern <stern@rowland.harvard.edu>]

On Thu, 22 Nov 2012, naveen yadav wrote:

> Hi,
> 
> we are using 3.2.2 version of kernel.
> 
> When Bluetooth(connected to USB 2.0) is suspended, urb->use_count
> value does not become zero and eventually there is a continuous wait
> in wait_event() of usb_kill_urb().
>           wait_event(usb_kill_urb_queue, atomic_read(&urb->use_count) == 0);
> 
> urb_count is incremented at only one place in usb_hcd_submit_urb() and

That's how it is _supposed_ to work.  You have discovered that
something else increments urb->use_count.

> decremented in usb_hcd_giveback_urb(). (Also, in error case inside
> usb_hcd_submit_urb()). After taking print of  urb->use_count at all
> these places, I see, even without connecting the Bluetooth device
> urb_count is incremented while below execution has never been
> decremented.

> .....
> [    3.490000] ====>[usb_hcd_giveback_urb][1626] urb->use_count = 0
> pipe=2147484032 [vid=0x1d6b] [pid=0x0002][devno=1]
> [    3.540000] ====>[usb_hcd_submit_urb][1482] urb->use_count = 2
> pipe=1077969280 [vid=0x1d6b] [pid=0x0002][devno=1]
> [    3.540000] ====>[usb_hcd_giveback_urb][1626] urb->use_count = 1
> pipe=1077969280 [vid=0x1d6b] [pid=0x0002][devno=1]

Now you need to find out what other code has changed urb->use_count and
fix it.

Alan Stern