The signal state field in G_TUNER is typically scaled from 0-100%.  Since we
don't know the signal level, we really would prefer the field to contain 100%
than 1/256, which in many utilities (such as v4l2-ctl) rounds to 0% even when
a signal is actually present.

This patch makes the behavior consistent with other drivers.

Signed-off-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

Tweak the comb filter config when in s-video mode to match the Hauppauge
Windows driver values (based on register dumps).

This work was sponsored by GetWellNetwork Inc.

Signed-off-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

Straighten out the clamp control configurations for the various video modes,
based off of register diffs done against the Hauppauge Windows driver.

This work was sponsored by GetWellNetwork Inc.

Signed-off-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

Cleanup the routine which disables the audio decoder.  In particular, the i2s
output was being shut off but the audio processor itself was not, resulting
in more power being consumed than needed.

Signed-off-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

The chip's default value for the brightness didn't match what we were sending
back in the queryctrl ioctl(), so if the application actually set the
brightness to the "default", it would actually end up being way too bright.

This work was sponsored by GetWellNetwork Inc.

Signed-off-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

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

g/s_fmt is going to disappear, so if it is not doing anything, then
just remove it.

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

Address a problem found in MythTV where if we are in digital mode, switch to
analog mode, and the switch back to digital mode, the first tuning request
after switching back to digital mode gets dropped.  This is because the au8522
maintains internal state, and would think the demod was already tuned to the
target frequency.

Thanks to Zaphod Beeblebrox for reporting this issue.

Signed-off-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

With applications like MythTV, switching inputs results in closing the digital
side and then immediately opening the analog side.  This exposes a race
condition where the dvb_frontend kernel thread powers down the chip and closes
the i2c gate even though we're in the middle of bringing up the analog part
of the chip (since the shutdown of the dvb_frontend kernel thread occurs
asychronously).

Introduce a construct to keep track of what mode we're in, and drop requests
to power down or management the gate if we've already switched to analog mode.

Thanks to Zaphod Beeblebrox for reporting this issue.

Signed-off-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>

Make the local filter coefficients static and const. This will eliminate the
following sparse warnings (see "make C=1"):
 * au8522_decoder.c:71:31: warning: symbol 'filter_coef' was not declared. Should it be static?
 * au8522_decoder.c:113:31: warning: symbol 'lpfilter_coef' was not declared. Should it be static?

Signed-off-by: Márton Németh <nm127@freemail.hu>
Acked-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>