The units for DTV_FREQUENCY are kHz for satellital delivery systems
(DVB-S/DVB-S2/DVB-TURBO/ISDB-S). Fix it at the API spec.
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This fixes bug introduced by multi-frontend to single-frontend change.

Finally HAS_LOCK is got back!
We are not allowed to access hardware in sleep mode...
Chip did not like when .get_frontend() reads some registers while
chip was sleeping and due to that HAS_LOCK bit was never gained.

TODO: We should add logic for dvb-core to drop out illegal calls like that.
Signed-off-by: NAntti Palosaari <crope@iki.fi>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Fix yet another bug introduced be recent cxd2820r multi-frontend to
single-frontend change.

Finally, we have at least almost working picture for DVB-C too.
Signed-off-by: NAntti Palosaari <crope@iki.fi>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

As pointed out by 'make versioncheck', there's no need for
drivers/media/dvb/frontends/tda18271c2dd.c to
Signed-off-by: NJesper Juhl <jj@chaosbits.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Fix another bug introduced by recent multi-frontend to single-frontend
change.
Signed-off-by: NAntti Palosaari <crope@iki.fi>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

cxd2820r implements only one frontend currently which
handles all the standards.
Signed-off-by: NAntti Palosaari <crope@iki.fi>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The caller doesn't check the return value of check_firmware() but static
checkers complain.  It currently returns negative error codes, or zero
or greater on success but since the return type is boolean the values
are truncated to one or zero.  I've changed it to return an int,
negative on error and zero on success.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

These were initialized twice by mistake.  They were defined the same way
both times so this doesn't change how the code works.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NMiroslav Slugen <thunder.mmm@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

In xc4000 chipsets real signal and noise level is stored in register
0x0A and 0x0B,so we can use those registers to monitor signal strength.

I tested this patch on 2 different cards Leadtek DVR3200 and DTV2000H
Plus, both with same results, I used special antenna hubs (toner 4x, 6x,
8x and 12x) with mesured signal lost, both registers are in dB value,
first represent signal with limit value -113.5dB (should be -114dB) and
exactly match with test results. Second represents noise level also in
dB and there is no maximum value, but from tests we can drop everything
above 32dB which tuner realy can't use, signal was usable till 20dB
noise level.

In digital mode we can take signal strength but sadly noise level is not
relevant and real value is stored in demodulator for now just zl10353,
also digital mode is just for testing, because it needs changing other
parts of code which reads data only from demodulator.

In analog mode I was able to test only FM radio, signal level is not
important, it says something about cable and hub losts, but nothing
about real quality of reception, so even if we have signal level at
minimum 113dB we can still here radio, because of that it is displaied
only in debug mode, but for real signal level is used noise register
which is again very accurate, radio noise level was betwen 6-20dB for
good signal, 20-25dB for medium signal, and above 25dB signal is
unusable.

For now real benefit of this patch is only for FM radio mode.
Signed-off-by: NMiroslav Slugen <thunder.mmm@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Update xc4000 tuner definition, number 81 is already in use by
TUNER_PARTSNIC_PTI_5NF05.

Cc: stable@kernel.org
Signed-off-by: NMiroslav Slugen <thunder.mmm@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

All radio tuners in cx88 driver using same address for radio and tuner,
so there is no need to probe it twice for same tuner and we can use
radio_type UNSET, this also fix broken radio since kernel 2.6.39-rc1
for those tuners.

Cc: stable@kernel.org
Signed-off-by: NMiroslav Slugen <thunder.mmm@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

All radio tuners in cx23885 driver using same address for radio and
tuner, so there is no need to probe it twice for same tuner and we can
use radio_type UNSET.

Be aware radio support in cx23885 is not yet committed, so this is only
minor fix for future support.
Signed-off-by: NMiroslav Slugen <thunder.mmm@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Fix possible null dereference for Leadtek DTV 3200H
XC4000 tuner when no firmware file available.

Cc: stable@kernel.org
Signed-off-by: NMiroslav Slugen <thunder.mmm@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

"dev" is NULL here so we should use "nr" instead of "dev->devno".
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Clearly ">=" was intended here instead of ">".
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

This reduces our module init to a simple usb_register() call, so
that we can make use of the new upcoming macro's for this.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The code for this is rather crufty, and being able to tie a device
to a specific minor is not really something we want to support in
a modern udev based world.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The pwc driver used to:
1. kmalloc a buffer
2. memcpy data to send over usb there
3. do the usb_control_msg call (which does not work with data on the stack)
4. free the buffer

For every usb command send. This patch changes the code to instead malloc
a buffer for this purpose once and use it everywhere.

[mchehab@redhat.com: Fix a compilation breakage with allyesconfig:
 drivers/media/video/pwc/pwc-ctrl.c: In function ‘pwc_get_cmos_sensor’:
 drivers/media/video/pwc/pwc-ctrl.c:546:3: warning: passing argument 4 of ‘recv_control_msg’ makes integer from pointer without a cast [en$
 drivers/media/video/pwc/pwc-ctrl.c:107:12: note: expected ‘int’ but argument is of type ‘unsigned char *’
 drivers/media/video/pwc/pwc-ctrl.c:546:3: error: too many arguments to function ‘recv_control_msg’
 drivers/media/video/pwc/pwc-ctrl.c:107:12: note: declared here]
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Since we always do a set_video_mode on stream start, there is no need
to actually send the mode info to the device on a s_fmt / s_parm ioctl.
Not doing this saves us doing (slow) usb io.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Before this patch various code in the mode setting patch checked
pdev->pixfmt, but that was not set until the mode setting succeeded, so
it was looking at the old pixfmt! This patch fixes this by making the
pixfmt a parameter to set_video_mode, and setting it from set_video_mode
on success.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Instead of messing around with id's it's much easier to just compare
against a filehandle pointer.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

For some reason the cx18 driver could open the radio device only once.
Remove this limitation.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

For some reason the /dev/radio device was implemented as an exclusive open:
you could open it only once and not a second time.

Remove this limitation.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Acked-by: NAndy Walls <awalls@md.metrocast.net>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

As per the feature removal document, make the tuner type check more strict
so that it is no longer possible to set the radio frequency through a video
node or the TV frequency through a radio node.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

As Rupert pointed out, the phrase "It is good practice" should be replaced
with "You must".
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Cc: Rupert Eibauer <Rupert.Eibauer@ces.ch>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Update the spec to the behavior implemented by the control framework.
This should have been documented long ago but for some reason it was
never done.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The omap_vout driver has an output overlay, but never advertised that
capability.

The driver should also set the V4L2_FBUF_FLAG_OVERLAY flag.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
CC: Archit Taneja <archit@ti.com>
CC: Vaibhav Hiremath <hvaibhav@ti.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The zoran driver does not support this flag, so don't set it.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The error handling in the original code wasn't complete so static
checkers complained about a potential NULL deference.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

The changeset 240ab508 is incomplete, as the first thing that
happens at cache clear is to do a memset with 0 to the cache.

So, the delivery system needs to be explicitly preserved there.
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

As reported by Lawrence[1], MythTV 0.24.1 does the wrong thing
with a DVBv5 call: it fills the delivery system with
SYS_UNDEFINED, expecting that the DVB core would work with that.

This used to work by accident, as the DVB core were missing the
check for the supported delivery systems. Yet, fixing it
is easy, so let's add a logic to handle this case, to
provide backward compatibility.

[1] http://patchwork.linuxtv.org/patch/8314/Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

I added it by mistake. It is useless as no real hardware.
It even uses even Cypress FX2, general USB bridge chip, default IDs
that makes driver load all FX2 devices having default ID...
Signed-off-by: NAntti Palosaari <crope@iki.fi>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

Some time ago, Manoel sent us a patch adding more stuff
to the init sequence. However, his patch were also doing
non-related stuff, by changing the init logic without
any good reason. So, it was asked for him to submit a
patch with just the data that has changed, in order to
allow us to better analyze it.

As he didn't what it was requested, I finally found some
time to dig into his init sequence and add it here.

Basically, new stuff is added there. There are a few changes:

1) The removal of the extra (duplicated) logic that puts
   the chip into the serial mode;
2) Some Viterbi VBER measurement init data was changed from
   0x00 to 0xff for layer A, to match what was done for
   layers B and C.

None of those caused any regressions and both make sense
on my eyes.

The other parameters additions actually increased the
tuning quality for some channels. Yet, some channels that
were previously discovered with scan disappered, while
others appeared instead. This were tested in Brasilia,
with an external antena.

At the overall, it is now a little better. So, better to
add these, and then try to figure out a configuration that
would get even better scanning results.
Reported-by: NManoel Pinheiro <pinusdtv@hotmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>

On mb86a20s, some registers have sub-addresses, while others not.
In order to make easier to compare different settings, group them.

No functional changes in this patch.
Signed-off-by: NMauro Carvalho Chehab <mchehab@redhat.com>