The cobalt device is a PCIe card with 4 HDMI inputs (adv7604) and a
connector that can be used to hook up an adv7511 transmitter or an
adv7842 receiver daughterboard.

This device is used within Cisco but is sadly not available outside
of Cisco. Nevertheless it is a very interesting driver that can serve
as an example of how to support HDMI hardware and how to use the popular
adv devices.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This makes it easier for the bridge driver to just passthrough such
events to the corresponding device node.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

All drivers use custom notifications, in particular when source changes.
The bridge only has to map the subdev that sends it to whatever video node it is connected to.
Signed-off-by: NJean-Michel Hautbois <jean-michel.hautbois@vodalys.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

vb2_buffer_done() already allows STATE_QUEUED, but currently only when not
streaming. It is useful to allow it while streaming as well, as this makes
it possible for drivers to requeue buffers while waiting for a stable
video signal.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Replace the dummy video format operations by pad format operations that
configure the output format.

Copied from the adv7604 driver.

Note: while arch/blackfin/mach-bf609/boards/ezkit.c uses adv7842_platform_data
this source has not been updated because it is broken since the very
beginning. It depends on a struct adv7842_output_format that does not
exist.

And besides that gcc has no support for bf609 so nobody can compile it
except by installing a toolchain from ADI.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Cc: Scott Jiang <scott.jiang.linux@gmail.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Remove multiple blank likes.
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

These state vars are used for more than just ber calculation, also
fix typo.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

BER & UCB aren't available unless we're locked; don't update dvbv5
stats when not locked and mark these counters as unavailable.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

DVBv3 is a legacy API. Drivers should use DVBv5, in order to support
modern applications. So, implement UCB using dvbv5.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Instead of reading BER again for DVBv3 call, use the value from the cache.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

DVBv3 is a legacy API. Drivers should use DVBv5, in order to support
modern applications. So, implement BER using dvbv5.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Use const on the static data, as gcc may optimize better the
code. Also, would prevent that some code would override the
data there.
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>

drivers/media/dvb-frontends/cx24120.c:1182:5: warning: no previous prototype for 'cx24120_init' [-Wmissing-prototypes]
 int cx24120_init(struct dvb_frontend *fe)
     ^
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>

As reported by smatch:
drivers/media/dvb-frontends/cx24120.c: In function 'cx24120_message_send':
drivers/media/dvb-frontends/cx24120.c:368:6: warning: variable 'ret' set but not used [-Wunused-but-set-variable]
  int ret, ficus;
      ^

The values written by cx24120 are never checked. So, remove the
check here too. That's said, the best would be to do the reverse,
but globally: to properly handle the error codes.
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>

Add in all the likely values for setting fec on tuning. Some values
yet to be tested but very likely to be correct.
In the very unlikely event that these are wrong tuning with
FEC_AUTO will still work.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

If set, all the other drivers reset no_base_addr on failure to
attach - git commit for this being added to some of them seems to
be eccd15aa

This driver has been floating around outside the mainline for so
long it hasn't had this fix, so add it in.

Whilst here tidy surrounding code to fix style issues.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Remove reg1 by refactoring for loops
Change ret into an int as it's only used for return values now,
and remove reset_result by using ret instead.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Use reg instead of ret for all calls to readreg, remove ret_EA
as it doesn't need to be a separate var anymore.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Probably a remnant of this driver being reverse engineered,
cx24120_init assigned ret on each call to writereg - they're not
used for anything so remove them to clear up the codebase.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Remove __func__ from calls to dev_dbg as dynamic debug can add in
the function name anyway.
Remove debug call in dvbv3 read_signal_strength as userspace has
this value anyway.
Reword some strings to make them simpler / more obvious.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Better to report CNR via DVBv5 stats, as scale can be used.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Previous version of this driver had a memset before every call
to cmd. This meant the default value of cmd.arg[1] was zero unless
burst is set. Make sure it remains zero.
Also fe_sec_mini_cmd_t is an enum, so test against it.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Algorithm copied from cx24117.c - it seems to work. Showing 95% on
a signal I know to be somewhere around 74dBuV.

[mchehab@osg.samsung.net: merged two patches, to avoid breaking compilation]
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This patch changes DVBv3 signal strength to read from the cache by
moving the signal strength reading routine into get_stat, and
reducing read_signal_strength to just returning the cached value.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

get_frontend was making a call to read_status, which is being
called periodically anyway. Have read_status store fe_status in
the state struct instead and use that within get_frontend.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Previous patch added in collection, this patch enables them by
signalling they are available.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Add new get_stats function, called from read_status, for
collecting DVBv5 stats into the frontend cache. Only signal
strength for now, can add others later.
Not currently marked as available, future patch will enable.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Change calculate_vco function to send_vco to tidy up cx24120_init
function.
xxyyzz variable is remnants of a manual div & remainder codepath
to perform a u64 rounded div; replace with kernel macro
DIV_ROUND_CLOSEST_ULL
Hex values provided to the message are mainly variables calculated
within this function, replace with these to remove hardcoding.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

msg.len has been ++'d so msg.len is one too many.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This patch adds in a missing command to cx24120_check_cmd. This part
of the old reverse engineered driver was quite hard to follow - it's
presumed the compiler did some neat optimisation that wasn't easy to
decode. This command was checked for in that version but not copied
across to the new switch statement.

For reference, old reverse engineered code:
xxzz = cmd->id - 0x11;			// look for specific message id
if ( xxzz <= 0x13 ) {
	msg_cmd_mask = 1 << xxzz;
	//0x0F8021 // if cmd_id 17 or 22 or 33-36, 42, 47, 57-61 etc. disable mpeg output
	if ( msg_cmd_mask & 0x0F8021 ) {	// 000011111000000000100001b
		cx24120_msg_mpeg_output_global_config(state, 0);
		msleep(100);
		state->dvb_s2_mode &=  0xFD;	// reset mpeg out enable flag
	}
}
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This patch removes the requirement of waiting for tune in
set_frontend by moving the clock set call into read_status
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This patch changes DVB-S/DVB-S2 to the more commonly used
DVB-S/S2 in the description for DVB_CX24120
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This patches fixes all checkpatch-errors.
Even the ones created --strict.
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Mauro pointed out some coding style issues in his review.
This patch fixes them.
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Add maintainer for the cx24120 frontend-driver.
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Some i2c-hosts are quite limited regarding maximum
i2c-burst-write-sizes. This patch makes the previously
hardcoded field configurable by users of the driver.
Signed-off-by: NPatrick Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This patch fixes printk-related checkpatch warnings. All printks
have been converted to pr_warn or pr_info or pr_err.
Signed-off-by: NPatrick.Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

This patch adds support for the Technisat Skystar S2 - this
has been tried before but the cx24120 driver was a bit out of shape
and it didn't got any further:

https://patchwork.linuxtv.org/patch/10575/

It is an old card, but currently being sold off for next to nothing,
so it's proving quite popular of late. Noticing it's quite similar
to the cx24116 and cx24117 I've rewritten the driver in a similar way.

There were a few registers and commands from those drivers
missing from this one I've tested out and found they do something so
they've been added in to speed up tuning and to make get_frontend
return something useful.
Signed-off-by: NJemma Denson <jdenson@gmail.com>
Signed-off-by: NPatrick.Boettcher <patrick.boettcher@posteo.de>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>

Missing varlistentry tags.
Signed-off-by: NHans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: NMauro Carvalho Chehab <mchehab@osg.samsung.com>