The pvrusb2 driver was getting had by this scenario:

1. Task A calls kthread_stop() for task B.
2. Before exiting, then Task B calls kthread_stop() for task C.

The problem is, kthread_stop() wants to allocate an internal resource
to itself (i.e. acquire a lock), which won't be released until
kthread_stop() returns.  But kthread_stop() won't return until task B
is dead.  But task B won't die until it finishes its call to
kthread_stop() for task C, and that will block waiting on the resource
already allocated inside task A.  Deadlock.

With the pvrusb2 driver, task A is the caller to pvr_exit(), task B is
the control thread run inside of pvrusb2-context.c, and task C is any
worker thread run inside of pvrusb2-hdw.c.

This problem got introduced by the previous threading setup change,
which was itself an attempt to fix a module tear-down race (which it
actually did fix).  The lesson here is that a task being waited on as
part of a kthread_stop() simply cannot be allow to also issue a
kthread_stop() - or we make sure not to issue the enclosing
kthread_stop() until we know that the inner kthread_stop() has
completed first.  The solution for the pvrusb2 driver is some hackish
code which changes the main control thread tear down into a two step
process.  This then makes it possible to delay issuing the
kthread_stop() on the control thread until after we know that
everything has been torn down first.  (And yes, we really need that
kthread_stop() because it's the only way to safely guarantee that a
module-referencing kernel thread has safely returned back out of the
module before we finally remove the module.)
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

The pvrusb2 driver - for basically forever - was not enforcing a
proper module tear-down.  Kernel threads are used inside the driver
and all must be gone before the module can be safely removed.  This
changeset reimplements a chunk of pvrusb2-context.c to enforce this
correctly.  Unfortunately this is not a simple fix.  The new
implementation also cuts back on kernel thread usage; instead of there
being 1 control thread per instance now it's just 1 control thread
shared by all instances.  (By dropping to a single thread then the
module exit function can block on its shutdown and the thread itself
can monitor and cleanly shut down all of the other instances first.)
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

This change significantly rearranges pvr2_context level initialization
and operation:

1. A new kernel thread is set up for management of the context.

2. Destruction of the pvr2_context instance is moved into the kernel
   thread.  No other context is able to remove the instance; doing
   this simplifies lock handling.

3. The callback into pvrusb2-main, which is used to trigger
   initialization of each interface, is now issued from this kernel
   thread.  Previously it had been indirectly issued out of the work
   queue thread in pvr2_hdw, which led to deadlock issues if the
   interface needed to change a control setting (which in turn
   requires dispatch of another work queue entry).

4. Callbacks into the interfaces (via the pvr2_channel structure) are
   now issued strictly from this thread.  The net result of this is
   that such callback functions can now also safely operate driver
   controls without deadlocking the work queue.  (At the moment this
   is not actually a problem, but I'm anticipating issues with this in
   the future).

5. There is no longer any need for anyone to enter / exit the
   pvr2_context structure.  Implementation of the kernel thread here
   allows this all to be internal now, simplifying other logic.

6. A very very longstanding issue involving a mutex deadlock between
   the pvrusb2 driver and v4l should now be solved.  The deadlock
   involved the pvr2_context mutex and a globals-protecting mutex in
   v4l.  During initialization the driver would take the pvr2_context
   mutex first then the v4l2 interface would register with v4l and
   implicitly take the v4l mutex.  Later when v4l would call back into
   the driver, the two mutexes could possibly be taken in the opposite
   order, a situation that can lead to deadlock.  In practice this
   really wasn't an issue unless a v4l app tried to start VERY early
   after the driver appeared.  However it still needed to be solved,
   and with the use of the kernel thread relieving need for
   pvr2_context mutex, the problem should be finally solved.
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

There is a callback that is issued to into pvr2_context from pvr2_hdw
after initialization is done.  There was a probability that this
callback could get missed.  Fixed.
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

None of these files use any of the functionality promised by
asm/semaphore.h.  It's possible that they rely on it dragging in some
unrelated header file, but I can't build all these files, so we'll have
fix any build failures as they come up.
Signed-off-by: NMatthew Wilcox <willy@linux.intel.com>

This is a new implementation for video pipeline control within the
pvrusb2 driver.  Actual start/stop of the pipeline is moved to the
driver's kernel thread.  Pipeline stages are controlled autonomously
based on surrounding pipeline or application control state.  Kernel
thread management is also cleaned up and moved into the internal
control structure of the driver, solving a set up / tear down race
along the way.  Better failure recovery is implemented with this new
control strategy.  Also with this change comes better control of the
cx23416 encoder, building on additional information learned about the
peculiarities of controlling this part (this information was the
original trigger for this rework).  With this change, overall encoder
stability should be considerably improved.  Yes, this is a large
change for this driver, but due to the nature of the feature being
worked on, the changes are fairly pervasive and would be difficult to
break into smaller pieces with any semblence of step-wise stability.
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

Trying to temporarily check that the stream is not claimed during open
of the radio device is at best a race condition.  What's to stop
another app from claiming the stream anyway the instant after the
check is done?  The implementation for this was dicey anyway.  So it's
removed.  The only "price" for this is that if /dev/radioX is opened
while streaming video, then the video stream is just going to switch
to radio mode anyway.  If a user does this, he gets what he expects...
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

Add (and expose) a new function, pvr2_channel_check_stream_no_lock(), in
pvrusb2-context.c. This is hopefully the last V4L2 interface related patch
to change anything outside pvrusb2-v4l2.c.

We need this to implement the open() for the radio device. The reason is
that within the *enter_context() section of open() we need to ensure nobody
is streaming and if we cannot, we should cleanup after ourselves and return
 -EBUSY. We cannot just use claim_stream() because:

   1) That would cause a deadlock trying to re-acquire the context lock
   2) We only need to ensure that nobody is streaming. We don't need to
      actually acquire the stream.

Again, this is a kinda ugly patch. Feel free to improve.
Signed-off-by: NPantelis Koukousoulas <pakt223@freemail.gr>
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

Fix up for make allyesconfig.
Signed-Off-By: NDavid Howells <dhowells@redhat.com>

Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>

Implement V4L2 driver for the Hauppauge PVR USB2 TV tuner.

The Hauppauge PVR USB2 is a USB connected TV tuner with an embedded
cx23416 hardware MPEG2 encoder.  There are two major variants of this
device; this driver handles both.  Any V4L2 application which
understands MPEG2 video stream data should be able to work with this
device.
Signed-off-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@infradead.org>