media: uvcvideo: Avoid NULL pointer dereference at the end of streaming
authorSakari Ailus <sakari.ailus@linux.intel.com>
Wed, 30 Jan 2019 10:09:41 +0000 (05:09 -0500)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 23 Mar 2019 13:35:30 +0000 (14:35 +0100)
commit 9dd0627d8d62a7ddb001a75f63942d92b5336561 upstream.

The UVC video driver converts the timestamp from hardware specific unit
to one known by the kernel at the time when the buffer is dequeued. This
is fine in general, but the streamoff operation consists of the
following steps (among other things):

1. uvc_video_clock_cleanup --- the hardware clock sample array is
   released and the pointer to the array is set to NULL,

2. buffers in active state are returned to the user and

3. buf_finish callback is called on buffers that are prepared.
   buf_finish includes calling uvc_video_clock_update that accesses the
   hardware clock sample array.

The above is serialised by a queue specific mutex. Address the problem
by skipping the clock conversion if the hardware clock sample array is
already released.

Fixes: 9c0863b1cc48 ("[media] vb2: call buf_finish from __queue_cancel")

Reported-by: Chiranjeevi Rapolu <chiranjeevi.rapolu@intel.com>
Tested-by: Chiranjeevi Rapolu <chiranjeevi.rapolu@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/media/usb/uvc/uvc_video.c

index a6d8002918831eebb0e60549e0c68c0fa52e84a6..393371916381ba4c2a346b5e9cd358d5b9e7fcb0 100644 (file)
@@ -638,6 +638,14 @@ void uvc_video_clock_update(struct uvc_streaming *stream,
        if (!uvc_hw_timestamps_param)
                return;
 
+       /*
+        * We will get called from __vb2_queue_cancel() if there are buffers
+        * done but not dequeued by the user, but the sample array has already
+        * been released at that time. Just bail out in that case.
+        */
+       if (!clock->samples)
+               return;
+
        spin_lock_irqsave(&clock->lock, flags);
 
        if (clock->count < clock->size)