Fixes an egregious bug in mceusb driver, where the receiver was being
put into idle mode far sooner than it should have, thanks to storing a
timeout value that in us where it should be ns. Basically, the receiver
kept going into idle mode before a trailing space had been fully
received, which was causing problems for some protocols, most notably
manifesting as lirc userspace never receiving a trailing space for any
rc5 signals.
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Fix a bug in v4l2_device_unregister where the sd pointer can be dereferenced
after it was freed.

Normally the i2c adapter is removed before this function is called. Removing
the adapter will also unregister all subdevs on that adapter, so generally
v4l2_device_unregister has nothing to do. However, in the case of a platform
i2c bus that bus is generally not freed.

In that case, after freeing the i2c subdevice the code will fall into the
second block when it tests if the subdev is a SPI device. But by that time
the subdev is already freed and the kernel oopses.

The fix is trivial: continue with the loop after freeing the i2c or spi
subdevice.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Reported-by: NDaniel Drake <dsd@laptop.org>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Zeroing video_device.dev causes a memory leak if video_set_drvdata
was called before video_register_device was called. video_set_drvdata
calls dev_set_drvdata which allocates video_device.dev.p.

memsetting this will prevent freeing of that memory.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

After a detach zero the whole device state to ensure a clean slate
on the next attach.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

v4l2_queryctrl sets the step value based on the control type. That would
be fine if it used the control type stored in the V4L2 kernel control
object, not the one stored in the userspace ioctl structure that has
just been memset to 0. Fix this.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The later makes extensive use of structures defined in the former.
Signed-off-by: NLaurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

When queryctrl is called with a V4L2_CID_PRIVATE_BASE control ID, then
currently it is replaced by the real internal ID. This is not according to
the spec so keep the V4L2_CID_PRIVATE_BASE ID in this case.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

VIDIOC_S_CTRL did not check against read-only controls. Even worse, for
controls of type CTRL_CLASS it would cause a kernel oops since those controls
do not have a s_ctrl op.
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

When registering an IR Rx device with the I2C subsystem, provide more detailed
information about the IR device and default remote configuration for the IR
driver modules.

Also explicitly register any IR Tx device with the I2C subsystem.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Acked-by: NMike Isely <isely@pobox.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This reverts a portion of commit

	44243fc2

A commit for which I errantly recommended that defaults for I2C address
0x71 not be set by ir-kbd-i2c.c

The pvrusb2 and bttv drivers currently rely on ir-kbd-i2c setting
defaults for that address.  Until I can get those bridge drivers fixed
to properly send IR_i2c_init_data for boards with Zilog Z8 chips,
just add back the default settings for I2C address 0x71.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The ir_command() function is a do-nothing stub; remove it.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The driver is now structured so that it must handle an IR Tx unit
for a Z8 IR chip, or it refuses to handle that Z8 IR chip.  This
allows us to assume that ir->tx != NULL in a few places in the driver,
and also allows us to always report Tx is available to userspace.
Get rid of unneeded tests for ir->tx == NULL and always report that
Tx is available.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The IR Rx polling thread was originally a kernel_thread long ago,
and had only been minimally converted to a kthread.  This patch
finishes that conversion by

- cleaning up all the unneeded completions

- destroying the kthread properly by calling kthread_stop()

- changing lirc_thread() to test kthread_should_stop() just before
	every point where it may sleep

- reorganizing the lirc_thread() function so it uses fewer lines

- modifying the name of the kthread from "lirc_zilog" to
	"zilog-rx-i2c-N", so ps will show which kthread polls
	which Zilog Z8 IR unit.

Also some minor tweaks were made to logging emitted by the
ir_probe() function.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This patch is an extensive rework of the ir_probe() and ir_remove() functions.

It removes all the double binding and allocation problems on module load.

It removes almost all the memory leaks on module exit and on device
instantiation failure. Proper destruction of the Rx polling kthread still
needs investigation and more work, but it is no worse than it already was.

This rework also had side effects that include:

- encapsulation of the ir_devices[] array
- serialization of access to the ir_devices[] array
- semantic change of the module parameter "disable_rx" to "tx_only"

If tx_only is true, the module does not claim the i2c_client for the IR Rx
function, and only claims and handles the i2c_client for the IR Tx function.
This is a first step in providing the option of letting ir-kbd-i2c.c handle
IR Rx function, while lirc_zilog handles the IR Tx function.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Don't make private copies of the i2c clients provided by the I2C
subsystem, don't change the client address field, and don't probe
the client addresses - the bridge driver already did that.  This
moves us to the proper I2C and binding model.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This change is a mostly mechanical break of the main struct IR
data structure into common, Rx, and Tx structures.  There were some
small logical changes required as well, such as eliminating "is_hdpvr",
to accomplish this.

This change is an intiial step in reworking lirc_zilog to decouple the
Rx and Tx handling as much as possible to fit with the new I2C
binding model.  This change actually makes lirc_zilog a little more
broken than it already was - memory deallocation in particular got worse.
However, this change makes the remaining problems easier to see and address.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The only reason to use the lirc_zilog module is for IR Tx, so remove
the possibility of disabling IR Tx.  If the user needs only IR Rx,
then the ir-kbd-i2c module works just fine, and doesn't require a
"firmware" image.
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Jean Delvare <khali@linux-fr.org> suggested this better
format for debug output in ir_probe().
Signed-off-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

After a recent update of xf86-input-evdev and xorg-server, I noticed
that X11 applications did not receive keypresses from the FireDTV
infrared remote control anymore.  Instead, the Xorg log featured lots of

    "FireDTV remote control: dropping event due to full queue!"

exclamations.  The Linux console did not have an issue with the
FireDTV's RC though.

The fix is to insert EV_SYN events after the key-down/-up events.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Rework device init part. Move common code part to a function.
Usefull for register multiple devices like video, radio, vbi etc.
Signed-off-by: NBeholder Intl. Ltd. Dmitry Belimov <d.belimov@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Commit e3c92215 ("[media] radio-aimslab.c: Fix
gcc 4.5+ bug") removed the include, but introduced new callers of msleep():

| drivers/media/radio/radio-aimslab.c: In function ‘rt_decvol’:
| drivers/media/radio/radio-aimslab.c:76: error: implicit declaration of function ‘msleep’
Signed-off-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Fix sparse warning for non-ANSI function declaration:

drivers/media/rc/ir-raw.c:247:30: warning: non-ANSI function declaration of function 'ir_raw_get_allowed_protocols'
Signed-off-by: NRandy Dunlap <randy.dunlap@oracle.com>
Cc:	Mauro Carvalho Chehab <mchehab@infradead.org>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This patch adds s_power support to adv7175 driver. Power-down is done
by power-down all four DACs.
Signed-off-by: NChristian Gmeiner <christian.gmeiner@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

info->num comes from the user.  It's type int.  If the user passes
in a negative value that would cause memory corruption.
Signed-off-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

My initial thinking was that we should default to mouse mode, so people
could use the mouse function to click on something on a login screen,
but a lot of systems where a remote is useful automatically log in a
user and launch a media center application, some of which hide the
mouse, which can be confusing to users if they punch buttons on the
remote and don't see any feedback. Plus, first and foremost, its a
remote, so lets default to being a remote, and only toggle into mouse
mode when the user explicitly asks for it. As a nice side-effect, this
actually simplifies some of the code a fair bit...
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The ene_ir driver was using a private define of MS_TO_NS, which is meant
to be microseconds to nanoseconds. The mceusb driver copied it,
intending to use is a milliseconds to microseconds. Lets move the
defines to a common location, expand and standardize them a touch, so
that we now have:

  MS_TO_NS - milliseconds to nanoseconds
  MS_TO_US - milliseconds to microseconds
  US_TO_NS - microseconds to nanoseconds
Reported-by: NDavid Härdeman <david@hardeman.nu>
CC: Maxim Levitsky <maximlevitsky@gmail.com>
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Otherwise, we have a null receive buffer, and the logic all falls down,
goes boom, all ffdc devs wind up as imon IR w/VFD. Oops.
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

dev->rdev is accessed in ene_setup_hw_settings, so it needs to be wired
up before then.

[Jarod Wilson]: Also fix a possible improper resource freeing bug while
we're looking at possible probe issues here.
Signed-off-by: NKyle McMartin <kmcmartin@redhat.com>
CC: Maxim Levitsky <maximlevitsky@gmail.com>
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

There's a nasty bug that slipped in when the rc device interface was
altered, only affecting the older 0xffdc imon devices. We were trying
to access ictx->rdev->allowed_protos before ictx->rdev had been set.

There's also an issue with call ordering that meant the correct
keymap wasn't getting loaded for MCE IR type 0xffdc devices.
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NThadeu Lima de Souza Cascardo <cascardo@holoscopio.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Use unlocked_ioctl in v4l2_file_operations. The locking is
already in place.
Signed-off-by: NMatti J. Aaltonen <matti.j.aaltonen@nokia.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

With cmwq, there's no reason to use separate out_work_queue.  Drop it
and use system_wq instead.  The in_work_queue needs to be ordered so
can't use one of the system wqs; however, as it isn't used to reclaim
memory, allocate the workqueue with alloc_ordered_workqueue() without
WQ_MEM_RECLAIM.
Signed-off-by: NTejun Heo <tj@kernel.org>
Reviewed-by: NAndy Walls <awalls@md.metrocast.net>
Acked-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Hi,

While reading
drivers/media/video/tlg2300/pd-video.c::alloc_bulk_urbs_generic() I
noticed that

 - We don't free the memory allocated to 'urb' if the call to
   usb_alloc_coherent() fails.
 - If the 'num' argument to the function is ever <= 0 we'll return an
   uninitialized variable 'i' to the caller.

The following patch addresses both of the above by a) calling
usb_free_urb() when usb_alloc_coherent() fails and by explicitly
initializing 'i' to zero.
I also moved the variables 'mem' and 'urb' inside the for loop. This does
not actually make any difference, it just seemed more correct to me to let
variables exist only in the innermost scope they are used.
Signed-off-by: NJesper Juhl <jj@chaosbits.net>
Acked-by: NHuang Shijie <shijie8@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

We may leak the storage allocated to 'state' in
drivers/media/dvb/frontends/ix2505v.c::ix2505v_attach() on error, as
it is too early to be able to call ix2505v_release().

This patch makes sure we free the allocated memory in the failure case.
Signed-off-by: NJesper Juhl <jj@chaosbits.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NJean-François Moine <moinejf@free.fr>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>