[media] xc5000: add IF output level control

Adds control of the IF output level to the xc5000 tuner
configuration structure.  Increases the IF level to the
demodulator to fix failure to lock and picture breakup
issues (with the au8522 demodulator, in the case of the
Hauppauge HVR950Q).

This patch works with all XC5000 firmware versions.

Signed-off-by: Richard Vollkommer <linux@hauppauge.com>
Signed-off-by: Michael Ira Krufky <mkrufky@linuxtv.org>
Reviewed-by: Devin Heitmueller <dheitmueller@kernellabs.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] lirc_igorplugusb: remove

This driver has been replaced by an rc-core driver for the same hardware.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] rc: port IgorPlug-USB to rc-core

This is a complete re-write inspired by the original lirc driver.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] m88ts2022: return the err code in its probe function when error occurs

if "chip_id" is wrong or "dev->cfg.clock_out" is invalid, the i2c model is still loaded.
It will cause "kernel NULL pointer dereference" oops when the i2c model remove.
returning the err code will prevent the i2c model load.

Signed-off-by: Nibble Max <nibble.max@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx23885: add DVBSky S950C and T980C RC support

DVBSky s950ci dvb-s/s2 ci PCIe card:
1>dvb frontend: M88TS2022(tuner),M88DS3103(demod)
2>ci controller: CIMAX SP2 or its clone.
3>PCIe bridge: CX23885

The patchs are based on the following patchs.
Olli Salonen submit:
https://patchwork.linuxtv.org/patch/26180/
https://patchwork.linuxtv.org/patch/26183/
https://patchwork.linuxtv.org/patch/26324/
Nibble Max submit:
https://patchwork.linuxtv.org/patch/26207/

Signed-off-by: Nibble Max <nibble.max@gmail.com>
Reviewed-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx23885: add DVBSky S950C dvb-s/s2 ci PCIe card support(no RC)

DVBSky s950ci dvb-s/s2 ci PCIe card:
1>dvb frontend: M88TS2022(tuner),M88DS3103(demod)
2>ci controller: CIMAX SP2 or its clone.
3>PCIe bridge: CX23885

The patchs are based on the following patchs.
Olli Salonen submit:
https://patchwork.linuxtv.org/patch/26180/
https://patchwork.linuxtv.org/patch/26183/
https://patchwork.linuxtv.org/patch/26324/
Nibble Max submit:
https://patchwork.linuxtv.org/patch/26207/

Signed-off-by: Nibble Max <nibble.max@gmail.com>
Reviewed-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] dvb-usb-dvbsky: add s960ci dvb-s/s2 usb ci box support

DVBSky s960ci dvb-s/s2 usb ci box:
1>dvb frontend: M88TS2022(tuner),M88DS3103(demod)
2>usb controller: CY7C86013A
3>ci controller: CIMAX SP2 or its clone.

Signed-off-by: Nibble Max <nibble.max@gmail.com>
Reviewed-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] Update Documentation cardlist

Some new devices got added. Update cardlists.

Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] em28xx: add support for Leadtek VC100 USB capture device

Leadtek VC100 is a simple USB capture stick, similar to
Yakumo Movie Mixer.

Signed-off-by: Witold Krecicki <wpk+lkml@culm.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] dvbsky: add option to disable IR receiver

Add an option to disable remote controller for DVBSky devices by specifying
the disable_rc option at modprobe.

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] dvbsky: clean logging

dev_err includes the function name in the log printout, so there is no need to include it manually. While here, fix a small grammatical error in the i2c error message.

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] dvbsky: don't print MAC address from read_mac_address

The dvb-usb-v2 already prints out the MAC address, no need to print
it out also here.

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: earthsoft: logging neatening

Use dev_err instead of pt1_printk
o reduce object code size
o remove now unused pt1_printk macro

Neaten dev_<level> uses in pt3
o add missing newlines
o align arguments
o remove unnecessary OOM messages as there's a generic one
o typo fixes in messages

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] dvb-frontends/Kconfig: better describe Toshiba TC90522

Toshiba TC90522 is wrongly categorized, fix it

Signed-off-by: ほち, AreMa Inc <knightrider@are.ma>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] Staging: media: lirc: cleaned up packet dump in 2 files

lirc_imon.c and lirc_sasem.c contain an incoming_packet method that
is using deprecated printk's.  Removed blocks replacing with single
dev_info with a %*ph format instead.

Signed-off-by: Amber Thrall <amber.rose.thrall@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: fix VBI support

Now works with both NTSC and PAL. Tested with CC/XDS for NTSC and
teletext/WSS for PAL. The start lines were wrong, the WSS signal
wasn't captured and there was no difference between NTSC and PAL
w.r.t. the count[] values so NTSC returned way too many lines.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: pci_disable_device comes after free_irq

Move pci_disable_device() down otherwise it will complain about an
unfreed irq.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: consistently use UNSET for absent tuner

Don't mix UNSET and TUNER_ABSENT: you have to pick one or the other. For
this driver selecting UNSET to represent an absent tuner resulting in
the fewest changes.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: don't allow changes while vb2_is_busy

Make sure that changing the standard or format is not allowed while
one or more of the video, vbi or mpeg vb2 queues are busy.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: drop mpeg_active field

The vb2 framework knows if streaming is in progress, no need to use
a separate field for that.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: move width, height and field to core struct

The width, height and field values are core fields since both vbi, video
and blackbird use the same video input.

Move those fields to the correct struct.

Also fix the field checks in the try_fmt functions: add V4L2_FIELD_SEQ_BT/TB
support and map incorrect field values to a correct field value instead of
returning an error.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: don't pollute the kernel log

There is no reason to dump the sram code to the kernel log when you
stop streaming. Remove those calls to cx88_sram_channel_dump.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: increase API command timeout

The timeout is way too small. Especially complicated command like
CX2341X_ENC_STOP_CAPTURE takes much more time than 10 ms. Increase the
timeout to 1 second, just as ivtv does (the cx88-blackbird has the
same IP core for MPEG compression as ivtv).

This solves a nasty issue where STOP_CAPTURE would timeout and the
mailbox is left in a busy state, making it impossible to start streaming
a second time without reloading the driver first.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: remove dependency on btcx-risc

btcx-risc is for the bt8xx driver and other drivers shouldn't depend
on it. There is no benefit to use that module just to do a
pci_zalloc_consistent.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: drop cx88_free_buffer

Remove this function. This makes all vb2 queues behave the same, which
simplifies comparing the various vb2 queue op implementations.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: return proper errors during fw load

Don't return -1, return a proper error.

Replace dprintk(0, ...) by pr_err since firmware load errors should just be
reported as an error.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: fix sparse warning

drivers/media/pci/cx88/cx88-blackbird.c:476:25: warning: cast to restricted __le32

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: convert to vb2

As usual, this patch is very large due to the fact that half a vb2 conversion
isn't possible. And since this affects blackbird, alsa, core, dvb, vbi and
video the changes are all over.

What made this more difficult was the peculiar way the risc program was setup.
The driver allowed for running out of buffers in which case the DMA would stop
and restart when the next buffer was queued. There was also a complicated
timeout system for when buffers weren't filled. This was replaced by a much
simpler scheme where there is always one buffer around and the DMA will just
cycle that buffer until a new buffer is queued. In that case the previous
buffer will be chained to the new buffer. An interrupt is generated at the
start of the new buffer telling the driver that the previous buffer can be
passed on to userspace.

Much simpler and more robust.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: drop videobuf abuse in cx88-alsa

The alsa driver uses videobuf low-level functions that are not
available in vb2, so replace them by driver-specific functions.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: drop the bogus 'queue' list in dmaqueue

This list is only used if the width, height and/or format of a buffer has
changed, but that can never happen. Remove it and all associated code.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx88: remove fmt from the buffer struct

This is a duplicate of dev->fmt and can be removed. As a consequence a
lot of tests that check if the format has changed midstream can be
removed as well: the format cannot change midstream, so this is a bogus
check.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: remove direct register PWR_CTL_EN modification that switches port3

The only remaining place that modifies the relevant bit is in function
cx231xx_set_Colibri_For_LowIF

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: scan all four existing i2c busses instead of the 3 masters

The scanning itself just fails (as before this series) but now
the correct busses are scanned.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: drop unconditional port3 switching

All switching should be done by i2c mux adapters.
Drop explicit dont_use_port_3 flag.
Drop info message about switch.

Only the removed code in start_streaming is questionable:
It did switch the port_3 flag without accessing i2c in between.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: register i2c mux adapters for bus 1

I2C bus 1 has internally a switch. Use it as I2C_1_MUX_1 and
I2C_1_MUX_3, letting the I2C core handling the switch.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: change usage of I2C_1 to the real i2c port

change almost all instances of I2C_1 to I2C_1_MUX_3

Only these cases are changed to I2C_1_MUX_1:
* All that have dont_use_port_3 set.
* CX231XX_BOARD_HAUPPAUGE_EXETER, old code did explicitly not switch to port3.
* eeprom access for 930C

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: let is_tuner check the real i2c port and not the i2c master number

Get used i2c port from bus_nr and status of port_3 switch.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: remember status of i2c port_3 switch

This is used later for is_tuner function that switches i2c behaviour for
some tuners.

[mchehab@osg.samsung.com: Fix CodingStyle on a multi-line comment]

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: add wrapper to get the i2c_adapter pointer

This is a preparation for mapping I2C_1_MUX_1 and I2C_1_MUX_3 later to the seperate
muxed i2c adapters.

Map mux adapters to I2C_1 for now.

Add local variables for i2c_adapters in dvb_init to get line lengths
shorter.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: Use symbolic constants for i2c ports instead of numbers

Replace numbers by the constants of same value and same meaning.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: Modifiy the symbolic constants for i2c ports and describe

Change to I2C_0 ... I2C_2 for the master ports
and add I2C_1_MUX_1 and I2C_1_MUX_3 for the muxed ones.

V2: Renamed mux adapters to seperate them from master adapters.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: give each master i2c bus a seperate name

Instead of using the same name for all 3 i2c physical buses
inside cx231xx, name them differently, adding a number
to it. This helps to better deal with the logs.

[mchehab@osg.samsung.com: removed an unused bus_name var from
 the original patch]
Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: delete i2c_client per bus

For each i2c master there is a i2c_client allocated that could be
deleted now that its only two users have been changed to use their
own i2c_client.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: use own i2c_client for eeprom access

This is a preparation for deleting the otherwise useless i2c_clients
that are allocated for all the i2c master adapters.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx231xx: let i2c bus scanning use its own i2c_client

This is a preparation for deleting the otherwise useless i2c_clients
that are allocated for all the i2c master adapters.

Signed-off-by: Matthias Schwarzott <zzam@gentoo.org>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] rc-main: fix lockdep splash for rc-main

lockdep reports a potential circular dependecy deadlock when registering input device.

Unlock mutex rc_dev->lock prior to calling ir_raw_event_register to avoid the circular
dependency since that function also calls input_register_device and rc_open.

 ======================================================
 [ INFO: possible circular locking dependency detected ]
 3.17.0-rc7+ #24 Not tainted
 -------------------------------------------------------
 modprobe/647 is trying to acquire lock:
  (input_mutex){+.+.+.}, at: [<ffffffff812ed81c>] input_register_device+0x2ba/0x381

 but task is already holding lock:
  (ir_raw_handler_lock){+.+.+.}, at: [<ffffffff813186ed>] ir_raw_event_register+0x102/0x190

 which lock already depends on the new lock.

[cut text]

 other info that might help us debug this:

 Chain exists of:
   input_mutex --> &dev->lock --> ir_raw_handler_lock

  Possible unsafe locking scenario:

        CPU0                    CPU1
        ----                    ----
   lock(ir_raw_handler_lock);
                                lock(&dev->lock);
                                lock(ir_raw_handler_lock);
   lock(input_mutex);

  *** DEADLOCK ***

 4 locks held by modprobe/647:
  #0:  (&dev->mutex){......}, at: [<ffffffff812d19f3>] device_lock+0xf/0x11
  #1:  (&dev->mutex){......}, at: [<ffffffff812d19f3>] device_lock+0xf/0x11
  #2:  (&dev->lock){+.+.+.}, at: [<ffffffff81317fff>] rc_register_device+0x55d/0x58a
  #3:  (ir_raw_handler_lock){+.+.+.}, at: [<ffffffff813186ed>] ir_raw_event_register+0x102/0x190

 stack backtrace:
 CPU: 0 PID: 647 Comm: modprobe Not tainted 3.17.0-rc7+ #24

 Call Trace:
  [<ffffffff81489d6a>] dump_stack+0x46/0x58
  [<ffffffff81487699>] print_circular_bug+0x1f8/0x209
  [<ffffffff81074353>] __lock_acquire+0xb54/0xeda
  [<ffffffff81080f17>] ? console_unlock+0x34d/0x399
  [<ffffffff81074c01>] lock_acquire+0xd9/0x111
  [<ffffffff812ed81c>] ? input_register_device+0x2ba/0x381
  [<ffffffff8148e650>] mutex_lock_interruptible_nested+0x57/0x381
  [<ffffffff812ed81c>] ? input_register_device+0x2ba/0x381
  [<ffffffff81124e03>] ? kfree+0x7c/0x96
  [<ffffffff812ed81c>] ? input_register_device+0x2ba/0x381
  [<ffffffff81072531>] ? trace_hardirqs_on+0xd/0xf
  [<ffffffff812ed81c>] input_register_device+0x2ba/0x381
  [<ffffffff8131a537>] ir_mce_kbd_register+0x109/0x139
  [<ffffffff81318728>] ir_raw_event_register+0x13d/0x190
  [<ffffffff81317e40>] rc_register_device+0x39e/0x58a
  [<ffffffff81072531>] ? trace_hardirqs_on+0xd/0xf
  [<ffffffffa00cf2e3>] nvt_probe+0x5ad/0xd52 [nuvoton_cir]
  [<ffffffffa00ced36>] ? nvt_resume+0x80/0x80 [nuvoton_cir]
  [<ffffffff81296003>] pnp_device_probe+0x8c/0xa9
  [<ffffffff812d1b94>] ? driver_sysfs_add+0x6e/0x93
  [<ffffffff812d203a>] driver_probe_device+0xa1/0x1e3
  [<ffffffff812d217c>] ? driver_probe_device+0x1e3/0x1e3
  [<ffffffff812d21ca>] __driver_attach+0x4e/0x6f
  [<ffffffff812d075b>] bus_for_each_dev+0x5a/0x8c
  [<ffffffff812d1b24>] driver_attach+0x19/0x1b
  [<ffffffff812d1879>] bus_add_driver+0xf1/0x1d6
  [<ffffffff812d2817>] driver_register+0x87/0xbe
  [<ffffffffa0120000>] ? 0xffffffffa0120000
  [<ffffffff81295da4>] pnp_register_driver+0x1c/0x1e
  [<ffffffffa0120010>] nvt_init+0x10/0x1000 [nuvoton_cir]
  [<ffffffff8100030e>] do_one_initcall+0xea/0x18c
  [<ffffffff8111497f>] ? __vunmap+0x9d/0xc7
  [<ffffffff810a3ca1>] load_module+0x1c21/0x1f2c
  [<ffffffff810a0bce>] ? show_initstate+0x44/0x44
  [<ffffffff810a404e>] SyS_init_module+0xa2/0xb1
  [<ffffffff81490ed2>] system_call_fastpath+0x16/0x1b

Signed-off-by: Tomas Melin <tomas.melin@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] af9005: fix kernel panic on init if compiled without IR

This patches fixes an ancient bug in the dvb_usb_af9005 driver, which
has been reported at least in the following threads:
https://lkml.org/lkml/2009/2/4/350
https://lkml.org/lkml/2014/9/18/558

If the driver is compiled in without any IR support (neither
DVB_USB_AF9005_REMOTE nor custom symbols), the symbol_request calls in
af9005_usb_module_init() return pointers != NULL although the IR
symbols are not available.

This leads to the following oops:
...
[    8.529751] usbcore: registered new interface driver dvb_usb_af9005
[    8.531584] BUG: unable to handle kernel paging request at 02e00000
[    8.533385] IP: [<7d9d67c6>] af9005_usb_module_init+0x6b/0x9d
[    8.535613] *pde = 00000000
[    8.536416] Oops: 0000 [#1] PREEMPT PREEMPT DEBUG_PAGEALLOCDEBUG_PAGEALLOC
[    8.537863] CPU: 0 PID: 1 Comm: swapper Not tainted 3.15.0-rc6-00151-ga5c075c #1
[    8.539827] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.7.5-20140531_083030-gandalf 04/01/2014
[    8.541519] task: 89c9a670 ti: 89c9c000 task.ti: 89c9c000
[    8.541519] EIP: 0060:[<7d9d67c6>] EFLAGS: 00010206 CPU: 0
[    8.541519] EIP is at af9005_usb_module_init+0x6b/0x9d
[    8.541519] EAX: 02e00000 EBX: 00000000 ECX: 00000006 EDX: 00000000
[    8.541519] ESI: 00000000 EDI: 7da33ec8 EBP: 89c9df30 ESP: 89c9df2c
[    8.541519]  DS: 007b ES: 007b FS: 0000 GS: 00e0 SS: 0068
[    8.541519] CR0: 8005003b CR2: 02e00000 CR3: 05a54000 CR4: 00000690
[    8.541519] Stack:
[    8.541519]  7d9d675b 89c9df90 7d992a49 7d7d5914 89c9df4c 7be3a800 7d08c58c 8a4c3968
[    8.541519]  89c9df80 7be3a966 00000192 00000006 00000006 7d7d3ff4 8a4c397a 00000200
[    8.541519]  7d6b1280 8a4c3979 00000006 000009a6 7da32db8 b13eec81 00000006 000009a6
[    8.541519] Call Trace:
[    8.541519]  [<7d9d675b>] ? ttusb2_driver_init+0x16/0x16
[    8.541519]  [<7d992a49>] do_one_initcall+0x77/0x106
[    8.541519]  [<7be3a800>] ? parameqn+0x2/0x35
[    8.541519]  [<7be3a966>] ? parse_args+0x113/0x25c
[    8.541519]  [<7d992bc2>] kernel_init_freeable+0xea/0x167
[    8.541519]  [<7cf01070>] kernel_init+0x8/0xb8
[    8.541519]  [<7cf27ec0>] ret_from_kernel_thread+0x20/0x30
[    8.541519]  [<7cf01068>] ? rest_init+0x10c/0x10c
[    8.541519] Code: 08 c2 c7 05 44 ed f9 7d 00 00 e0 02 c7 05 40 ed f9 7d 00 00 e0 02 c7 05 3c ed f9 7d 00 00 e0 02 75 1f b8 00 00 e0 02 85 c0 74 16 <a1> 00 00 e0 02 c7 05 54 84 8e 7d 00 00 e0 02 a3 58 84 8e 7d eb
[    8.541519] EIP: [<7d9d67c6>] af9005_usb_module_init+0x6b/0x9d SS:ESP 0068:89c9df2c
[    8.541519] CR2: 0000000002e00000
[    8.541519] ---[ end trace 768b6faf51370fc7 ]---

The prefered fix would be to convert the whole IR code to use the kernel IR
infrastructure (which wasn't available at the time this driver had been created).

Until anyone who still has this old hardware steps up an does the conversion,
fix it by not calling the symbol_request calls if the driver is compiled in
without the default IR symbols (CONFIG_DVB_USB_AF9005_REMOTE).
Due to the IR related pointers beeing NULL by default, IR support will then be disabled.

The downside of this solution is, that it will no longer be possible to
compile custom IR symbols (not using CONFIG_DVB_USB_AF9005_REMOTE) in.

Please note that this patch has NOT been tested with all possible cases.
I don't have the hardware and could only verify that it fixes the reported
bug.

Reported-by: Fengguag Wu <fengguang.wu@intel.com>
Signed-off-by: Frank Schäfer <fschaefer.oss@googlemail.com>
Cc: <stable@vger.kernel.org>
Acked-by: Luca Olivetti <luca@ventoso.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx23885: add IR for DVBSky T9580 Dual DVB-S2/T2/C PCIe card

DVBSky T9580 uses Integrated CX23885 IR controller to decode IR signal.
The IR type of DVBSky remote control is RC5.

Signed-off-by: Nibble Max <nibble.max@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx23855: add CI support for DVBSky T980C

Add CI support for DVBSky T980C card. The new host device independent CIMaX SP2 I2C driver was used to implement it.

IRQ handling is not implemented at this point. It could be used to detect the CAM insertion/removal instantly.

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx23885: add I2C client for CI into state and handle unregistering

If the CI chip has an I2C driver, we need to store I2C client into state.

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] sp2: improve debug logging

Improve debugging output.

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] sp2: fix incorrect struct

Incorrect struct used in the SP2 driver.

Reported-by: Max Nibble <nibble.max@gmail.com>
Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] cx23855: add support for DVBSky T980C (no CI support)

This patch adds basic support for DVBSky T980C card. CI interface is not supported.

DVBSky T980C is a PCIe card with the following components:
- CX23885 PCIe bridge
- Si2168-A20 demodulator
- Si2158-A20 tuner
- CIMaX SP2 CI chip

The demodulator and tuner need firmware. They're the same as used with TT CT2-4650 CI:
https://www.mail-archive.com/linux-media@vger.kernel.org/msg78033.html

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Reviewed-by: Antti Palosaari <crope@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] ti-fpe: LLVMLinux: Remove nested function from ti-vpe

Replace the use of nested functions where a normal function will suffice.

Nested functions are not liked by upstream kernel developers in general. Their
use breaks the use of clang as a compiler, and doesn't make the code any
better.

This code now works for both gcc and clang.

Suggested-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Behan Webster <behanw@converseincode.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] sound: Update au0828 quirks table

The au0828 quirks table is currently not in sync with the au0828
media driver.

Syncronize it and put them on the same order as found at au0828
driver, as all the au0828 devices with analog TV need the
same quirks.

Cc: stable@vger.kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] sound: simplify au0828 quirk table

Add a macro to simplify au0828 quirk table. That makes easier
to check it against the USB IDs at drivers/media/usb/au0828/au0828-cards.c.

Cc: stable@vger.kernel.org
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] si2157: add support for SYS_DVBC_ANNEX_B

Set the property for delivery system also in case of SYS_DVBC_ANNEX_B.
This behaviour is observed in the sniffs taken with Hauppauge HVR-955Q
Windows driver.

Signed-off-by: Olli Salonen <olli.salonen@iki.fi>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] em28xx-dvb: remove unused mfe_sharing

This field is not used on this driver anymore. Remove it.

Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] drxk: Fix debug printks

This patch partially reverts 0fb220f2a5cb. What happened is
that the conversion of debug messages to use pr_debug() was a
bad idea, because one needing to debug would need to both
enable debug level via a modprobe parameter, and then to
enable the dynamic printk's.

So, for now, let's use printk() directly at dprintk().

Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] Documentation: FE_SET_PROPERTY requires R/W

FE_SET_PROPERTY requires to open the devnode on R/W mode, or
otherwise it will fail. Document it.

Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] dib7000p: get rid of an unused argument

dib7000p is for DVB, and not ISDB. So, there's no layer.

That removes this compilation warning:
drivers/media/dvb-frontends/dib7000p.c:1972: warning: 'i' is used uninitialized in this function

Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] v4l2-ctrls: fix sparse warning

The warning is simple:

drivers/media/v4l2-core/v4l2-ctrls.c:1685:15: warning: incorrect type in assignment (different address spaces)

but the fix isn't.

The core problem was that the conversion from user to kernelspace was
done at too low a level and that needed to be moved up. That made it possible
to drop pointers to v4l2_ext_control from set_ctrl and validate_new and
clean up this sparse warning because those functions now always operate
on kernelspace pointers.

Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: return -ENODATA for *dv_timings/*_std calls

this patch adds support for returning -ENODATA if the current
output doesn't support it.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: group v4l2_ioctl_ops

this patch groups the v4l2_ioctl_ops.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: drop unused member memory from vpbe_layer

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: use helpers provided by core if streaming is started

this patch uses vb2_is_busy() helper to check if streaming is
actually started, instead of driver managing it.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: add support for VIDIOC_EXPBUF

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: add support for VB2_DMABUF

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: use vb2_ioctl_* helpers

this patch adds support for using vb2_ioctl_* helpers.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: use fh handling provided by v4l

this patch converts the driver to use fh handling provided by the
v4l core instead of driver doing it.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: use vb2_fop_mmap/poll

this patch teaches vpbe driver to use vb2_fop_mmap/poll helpers.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: improve vpbe_buffer_prepare() callback

this patch improve vpbe_buffer_prepare() callback, as buf_prepare()
callback is never called with invalid state and check for
vb2_plane_vaddr(vb, 0) is dropped as payload check should
be done unconditionally.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: drop buf_cleanup() callback

this patch drops buf_cleanup() callback as this callback
is never called with buffer state active.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: use vb2_ops_wait_prepare/finish helper functions

this patch makes use of vb2_ops_wait_prepare/finish helper functions.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: drop buf_init() callback

this patch drops the buf_init() callback as init
of buf list is not required.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] media: davinci: vpbe: initialize vb2 queue and DMA context in probe

this patch moves the initialization of vb2 queue and the DMA
context to probe() and clean up in remove() callback respectively.

Signed-off-by: Lad, Prabhakar <prabhakar.csengg@gmail.com>
Signed-off-by: Hans Verkuil <hans.verkuil@cisco.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: correct the formats info for encoder

The NV12M is supported by all the version of MFC, so it is better
to use it as default OUTPUT format.
MFC v5 doesn't support NV21, I have tested it, for the SEC doc
it is not supported either.

Signed-off-by: ayaka <ayaka@soulik.info>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: re-queue buffers if start_streaming fails

Patch b906352c2338 ([media] coda: dequeue buffers if start_streaming fails)
incorrectly marked buffers as DEQUEUED in case of start_streaming failure, when
in fact they should be set to QUEUED. The videobuf2 core warns about this.

Reported-by: Jean-Michel Hautbois <jean-michel.hautbois@vodalys.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: fix a race in interrupt flags handling

Interrupt result flags have to be cleared before a hardware job is run.
Otherwise, if they are cleared asynchronously, we may end up clearing them
after the interrupt for which we wanted to wait has already arrived, thus
overwriting the job results that we intended to wait for.

To prevent this, clear the flags only under hw_lock and before running
a hardware job.

Signed-off-by: Pawel Osciak <posciak@chromium.org>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: fix V4L2_CID_MIN_BUFFERS_FOR_CAPTURE on resolution change

G_CTRL on V4L2_CID_MIN_BUFFERS_FOR_CAPTURE will fail if userspace happens to
query it after getting a resolution change event and before the codec has
a chance to parse the header and switch to an initialized state.

Signed-off-by: Pawel Osciak <posciak@chromium.org>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: Remove unused alloc field from private buffer struct

This field is no longer used as MFC driver doesn't use vb2 alloc contexts
anymore.

Signed-off-by: Pawel Osciak <posciak@chromium.org>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: flush dpbs when resolution changes

While resolution change is detected by MFC, we flush out
older dpbs, send the resolution change event to application,
and then accept further queuing of new src buffers.

Sometimes, we error out during dpb flush because of lack of src
buffers. Since we have not started decoding new resolution yet,
it is simpler to push zero-size buffer until we flush out all dpbs.

This is already been done while handling EOS command, and this patch
extends the same logic to resolution change as well.

Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: De-init MFC when watchdog kicks in

If the software watchdog kicks in, we need to de-init MFC
before reloading firmware and re-intializing it again.

Signed-off-by: Arun Mankuzhi <arun.m@samsung.com>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: modify mfc wakeup sequence for V8

MFC wakeup command has to be sent after the host receives
firmware load complete status from risc.

Signed-off-by: Arun Mankuzhi <arun.m@samsung.com>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: Don't crash the kernel if the watchdog kicks in

If the software watchdog kicks in, the watchdog worker is not synchronized
with hardware interrupts and does not block other instances. It's possible
for it to clear the hw_lock, making other instances trigger a BUG() on
hw_lock checks. Since it's not fatal to clear the hw_lock to zero twice,
just WARN in those cases for now. We should not explode, as firmware will
return errors as needed for other instances after it's reloaded, or they
will time out.

A clean fix should involve killing other instances when watchdog kicks in,
but requires a major redesign of locking in the driver.

Signed-off-by: Pawel Osciak <posciak@chromium.org>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: declare s5p_mfc_bus_reset as static

drivers/media/platform/s5p-mfc/s5p_mfc_ctrl.c:132:5: warning: no previous prototype for 's5p_mfc_bus_reset' [-Wmissing-prototypes]
 int s5p_mfc_bus_reset(struct s5p_mfc_dev *dev)
     ^

Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: check mfc bus ctrl before reset

during reset sequence, it is advisable to follow the below
sequence, in order to avoid unexpected behavior from MFC
. set SFR 0x7110 MFC_BUS_RESET_CTRL 0x1
  // wait for REQ_STATUS to be 1
. get SFR 0x7110 MFC_BUS_RESET_CTRL 0x3
  // reset now

Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: keep RISC ON during reset for V7/V8

Reset sequence for MFC V7 and V8 do not need RISC_ON
to be set to 0, while for MFC V6 it is still needed.

Also, remove a couple of register settings during Reset
which are not needed from V6 onwards.

Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: Only set timestamp/timecode for new frames

Timestamp i of a previously decoded buffer was overwritten for some
H.264 streams with timestamp i+1 of the next buffer. This happened when
encountering frame_type S5P_FIMV_DECODE_FRAME_SKIPPED, indicating no
new frame.

In most cases this wrong indexing might not have been noticed except
for a one frame delay in frame presentation. For H.264 streams though
that require reordering of frames for presentation, it caused a slightly
erratic presentation time lookup and consequently dropped frames in the
Pepper Flash plugin.

Signed-off-by: Ilja H. Friedel <ihf@google.com>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: clear 'enter_suspend' flag if suspend fails

The enter_suspend flag is set after we enter mfc suspend function but if
suspend fails after that due to any reason (like hardware timeout etc),
this flag must be cleared before returning an error. Otherwise, this
flag never gets cleared and the MFC suspend will always return an error
on subsequent tries. If clock off fails, disable hw_lock also.

Signed-off-by: Prathyush K <prathyush.k@samsung.com>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: Fix REQBUFS(0) for encoder

Handle REQBUFS(0) for CAPTURE queue as well. Also use the proper queue to call
it on for OUTPUT.

Signed-off-by: Pawel Osciak <posciak@chromium.org>
Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] s5p-mfc: support MIN_BUFFERS query for encoder

Add V4L2_CID_MIN_BUFFERS_FOR_OUTPUT query for encoder.
Once mfc encoder state is HEAD_PARSED, which is sequence
header produced, dpb_count is avaialable. Let user space
query this value.

Signed-off-by: Kiran AVND <avnd.kiran@samsung.com>
Signed-off-by: Arun Kumar K <arun.kk@samsung.com>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: drop JPEG buffers not framed by SOI and EOI markers

This patch adds a quick check for valid JPEG frames before feeding them into
the bitstream buffer: Frames that do not begin with the JPEG start of image
marker and end with the end of image marker are dropped.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: set bitstream end flag in coda_release

This should fix CODA crashes due to timeouts when stopping
the decoding process with SIGINT.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: allow userspace to set compressed buffer size in a certain range

For small frame sizes, allocating 1 MiB per compressed buffer is a waste of
space. On the other hand, incompressible 1080p data can produce JPEGs larger
than 1 MiB at higher quality settings. Allow userspace to set the compressed
buffer size and clamp the value to a sensible range.
Also set the initial sizeimage to a value inside the range allowed by try_fmt.
While at it, reduce the default image size to a maximum of 1920*1088 (otherwise
JPEG will default to 8k*8k and 96 MiB buffers).

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: try to only queue a single JPEG into the bitstream

With bitstream padding, it is possible to decode a single JPEG in the bitstream
immediately. This allows us to only ever queue a single JPEG into the bitstream
buffer, except to increase payload over 512 bytes or to back out of hold state.
This is a measure to decrease JPEG decoder latency.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: pad input stream for JPEG decoder

Before starting a PIC_RUN, pad the bitstream with 0xff until 256 bytes
past the next multiple of 256 bytes, if the buffer to be decoded is the
last buffer in the bitstream.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: store bitstream buffer position with buffer metadata

Storing the buffer position in the bitstream with the buffer metadata
allows to later use that information to drop metadata for skipped buffers
and to determine whether bitstream padding has to be applied.

This patch also renames struct coda_timestamp to struct coda_buffer_meta
to make clear that it contains more than only the buffer timestamp.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: add CODA7541 JPEG support

This patch adds JPEG encoding and decoding support for CODA7541,
using the BIT processor. Separate JPEG encoder and decoder video
devices are created due to different streaming behaviour and
different supported pixel formats.
The hardware can not change subsampling on the fly, but encode
and decode 4:2:2 subsampled JPEG images from and into this format.

The CODA7541 JPEG decoder uses the bitstream buffer and thus can run
without new buffers queued if there is a buffer in the bitstream.

Since there is no standard way to store the colorspace used in
JPEGs, and to make v4l2-compliance happy, the JPEG format always
reports V4L2_COLORSPACE_JPEG.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>

[media] coda: add JPEG register definitions for CODA7541

Add JPEG specific sequence initialization registers and bit definitions.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Kamil Debski <k.debski@samsung.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>