media: cx88-alsa: number of pages should be unsigned long

As reported by smatch:

drivers/media/pci/cx88/cx88-alsa.c:286 cx88_alsa_dma_init() warn: should 'nr_pages << 12' be a 64 bit type?

the number of patches should be unsigned long.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cx23885-alsa: number of pages should be unsigned long

As reported by smatch:

drivers/media//pci/cx23885/cx23885-alsa.c:83 cx23885_alsa_dma_init() warn: should 'nr_pages << 12' be a 64 bit type?

the number of patches should be unsigned long.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cx25821-alsa: number of pages should be unsigned long

As reported by smatch:

drivers/media/pci/cx25821/cx25821-alsa.c:146 cx25821_alsa_dma_init() warn: should 'nr_pages << 12' be a 64 bit type?

the number of patches should be unsigned long.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: videobuf-dma-sg: number of pages should be unsigned long

As reported by smatch:

drivers/media/v4l2-core/videobuf-dma-sg.c:245 videobuf_dma_init_kernel() warn: should 'nr_pages << 12' be a 64 bit type?

The printk should not be using %d for the number of pages.

After looking better, the real problem here is that the
number of pages should be long int.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: serial_ir: use the right type for a dma address

As warned by smatch:
drivers/media/rc/serial_ir.c:550 serial_ir_probe() warn: should '8 << ioshift' be a 64 bit type?

the "8" constant should be unsigned long.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: tda10021: avoid casts when using symbol_rate

The usage of castings and float point when checking for
the setup based at the symbol_rate cause those warnings
with smatch:

drivers/media/dvb-frontends/tda10021.c:153 tda10021_set_symbolrate() warn: unsigned 'symbolrate' is never less than zero.
drivers/media/dvb-frontends/tda10021.c:155 tda10021_set_symbolrate() warn: unsigned 'symbolrate' is never less than zero.
drivers/media/dvb-frontends/tda10021.c:157 tda10021_set_symbolrate() warn: unsigned 'symbolrate' is never less than zero.
drivers/media/dvb-frontends/tda10021.c:159 tda10021_set_symbolrate() warn: unsigned 'symbolrate' is never less than zero.

While the code should work with gcc, as it will evaluate the
values into a constant before compiling, other compilers
could do otherwise. So, get rid of float pointing math on it,
avoiding the need of doing typecasts.

While here, cleanup some coding style issues at the related
code.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: place extern venus_fw_debug on a header file

Sparse warns about this symbol:

drivers/media/platform/qcom/venus/hfi_venus.c:133:5:  warning: symbol 'venus_fw_debug' was not declared. Should it be static?

Because hfi_venus.c doesn't include a header file with the
extern. So, move it to core.h, with is included by both
hfi_venus.c and dbgfs.c.

This way, if something changes with it, warnings or errors
will be produced.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: wl128x: get rid of a potential spectre issue

As reported by smatch:
drivers/media/radio/wl128x/fmdrv_common.c:736 fm_irq_handle_rdsdata_getcmd_resp() warn: potential spectre issue 'rds_fmt.data.groupdatabuff.buff>

Address it by using array_index_nospec().

While here, reorder the linux/ includes.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: av7110_v4l: avoid a typecast

While smatch reports an issue there:

drivers/media/pci/ttpci/av7110_v4l.c:163 ves1820_set_tv_freq() warn: unsigned 'freq' is never less than zero.
drivers/media/pci/ttpci/av7110_v4l.c:165 ves1820_set_tv_freq() warn: unsigned 'freq' is never less than zero.

The logic is actually fine. Yet, removing the typecast
shuts up smatch and makes the code more readable.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: qt1010: fix usage of unititialized value

As pointed by smatch:

drivers/media/tuners/qt1010.c:239 qt1010_init_meas1() error: uninitialized symbol 'val2'.
drivers/media/tuners/qt1010.c:273 qt1010_init_meas2() error: uninitialized symbol 'val'.

The logic is ok, but it is hard for static analyzers
to parse it, as it depends on a value read in the middle
of a loop.

Also, it takes a while for humans to verify.

Re-write the first function to use a more direct way.

At the second one, I opted to just initialize the read var,
in order to shut up the report.

While here, address a few coding style issues at the
function code.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: tda10086: cleanup symbol_rate setting logic

The original logic had an integer to unsigned integer
conversion, plus a float-point math. While gcc should be
able to do the match at compile time, other compilers might
not do the same. Also, those produce the following warnings
with static code analyzers:

drivers/media/dvb-frontends/tda10086.c:300 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:303 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:306 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:309 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:312 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:315 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:318 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:321 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:324 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.
drivers/media/dvb-frontends/tda10086.c:327 tda10086_set_symbol_rate() warn: unsigned 'symbol_rate' is never less than zero.

Change the logic to declare the reference constant as unsigned
and to not use float point math.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: fix "dubious: !x | !y" sparse warning

Upon running Sparse, "warning: dubious: !x | !y" is brought to notice
for this file.  Logical and bitwise OR are basically the same in this
context, so it doesn't cause a runtime bug.  But let's change it to
logical OR to make it a little bit cleaner and silence the Sparse
warning.

Signed-off-by: Anant Thazhemadam <anant.thazhemadam@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: Avoid comma separated statements

Use semicolons and braces.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: fix memleak in ia_css_stream_create

When aspect_ratio_crop_init() fails, curr_stream needs
to be freed just like what we've done in the following
error paths. However, current code is returning directly
and ends up leaking memory.

Signed-off-by: Dinghao Liu <dinghao.liu@zju.edu.cn>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: fix spelling mistake "unsupport" -> "unsupported"

There are spelling mistakes in some dev_err messages. Fix these.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: move null check to earlier point

`find_gmin_subdev()` that returns a pointer to `struct
gmin_subdev` can return NULL.

In `gmin_v2p8_ctrl()` there's a call to this function but the
possibility of a NULL was not checked before its being dereferenced,
i.e.:

  /* Acquired here --------v */
  struct gmin_subdev *gs = find_gmin_subdev(subdev);

  /*  v------Dereferenced here */
  if (gs->v2p8_gpio >= 0) {
      ...
  }

With this change we're null checking `find_gmin_subdev()` result
and we return an error if that's the case. We also WARN()
for the sake of debugging.

Signed-off-by: Cengiz Can <cengiz@kernel.wtf>
Reported-by: Coverity Static Analyzer CID 1465536
Suggested-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: mt9m114: replace fixed function names

There are a couple of debug messages using hardcoded function names
instead of the preferred __func__ magic constant.

Replace them:

WARNING: Prefer using '"%s...", __func__' to using 'misensor_rmw_reg', this function's name, in a string
215: FILE: ./media/atomisp/i2c/atomisp-mt9m114.c:215:
+       v4l2_err(client, "misensor_rmw_reg error exit, read failed\n");

WARNING: Prefer using '"%s...", __func__' to using 'misensor_rmw_reg', this function's name, in a string
236: FILE: ./media/atomisp/i2c/atomisp-mt9m114.c:236:
+       v4l2_err(client, "misensor_rmw_reg error exit, write failed\n");

Signed-off-by: Juan Antonio Aldea-Armenteros <juant.aldea@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: fix code style issues

Fix warnings and errors generated after running checkpatch.pl on pci/sh_css_firmware.c.

Signed-off-by: Rohit K Bharadwaj <bharadwaj.rohit8@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: Replace deprecated MSI APIs

Replace deprecated MSI IRQ enabler and disabler
with pci_alloc_irq_vectors and pci_free_irq_vectors respectively.
And as a result handle the returned error as appropriate.
Compile tested.

[mchehab: solved a merge conflict that renamed dev->pdev]

Signed-off-by: Suraj Upadhyay <usuraj35@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: fix identation at I2C Kconfig menu

There are several bad whitespacing usage there. Remove them.

While here, place all Kconfig options for sensors at the
same place.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: fix a brace coding sytle issue

Removed braces in 'if else' condition as it only consists of
one line each and according to coding style rules , in this case
the braces are not required.

Signed-off-by: Ankit Baluni <b18007@students.iitmandi.ac.in>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: Fix braces placement coding style errors

Fix braces placement errors as reported by checkpatch.pl

Signed-off-by: Aditya Jain <aditya.jainadityajain.jain@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: fixed a brace coding sytle issue

Removed braces for a 'if' condition as it contain only single line &
there is no need for braces for such case according to coding style
rules.

Signed-off-by: Ankit Baluni <b18007@students.iitmandi.ac.in>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: coding style: correct multiline comments

This fixes some coding style issues of multiline comments to
correct a few checkpatch.pl warnings.

Signed-off-by: Rahul Gottipati <rahul.blr97@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: atomisp: coding style: remove beginning whitespaces

This removes whitespaces at the beginning of a few lines to fix
some checkpatch.pl warnings.

Signed-off-by: Rahul Gottipati <rahul.blr97@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: open.rst: document mc-centric and video-node-centric

When we added support for omap3, back in 2010, we added a new
type of V4L2 devices that aren't fully controlled via the V4L2
device node.

Yet, we have never clearly documented in the V4L2 specification
the differences between the two types.

Let's document them based on the the current implementation.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: docs: add glossary.rst with common terms used at V4L2 spec

Add a glossary of terms used within the media userspace API
documentation, as several concepts are complex enough to cause
misunderstandings.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: open.rst: remove the minor number range

minor numbers use to range between 0 to 255, but that
was changed a long time ago. While it still applies when
CONFIG_VIDEO_FIXED_MINOR_RANGES, when the minor number is
dynamically allocated, this may not be true. In any case,
this is not relevant, as udev will take care of it.

So, remove this useless misinformation.

Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: open.rst: better document device node naming

Right now, only kAPI documentation describes the device naming.
However, such description is needed at the uAPI too. Add it,
and describe how to get an unique identifier for a given device.

Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
Reviewed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: docs: use SPDX GPL-2.0 OR GFDL-1.1 instead of text on *.svg files

There are some files that are dual licensed GPL and GFDL.

As SPDX v3.10 gained support for GFDL-1.1 with no invariant sections:

https://spdx.org/licenses/GFDL-1.1-no-invariants-or-later.html

Let's remove the dual license text, replacing them by:

SPDX-License-Identifier: GPL-2.0 OR GFDL-1.1-no-invariants-or-later

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: docs: use SPDX GPL-2.0 OR GFDL-1.1 instead of text on *.rst files

There are some files that are dual licensed GPL and GFDL.

As SPDX v3.10 gained support for GFDL-1.1 with no invariant sections:

https://spdx.org/licenses/GFDL-1.1-no-invariants-or-later.html

Let's remove the dual license text, replacing them by:

SPDX-License-Identifier: GPL-2.0 OR GFDL-1.1-no-invariants-or-later

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: docs: use the new SPDX header for GFDL-1.1 on *.svg files

SPDX v3.10 gained support for GFDL-1.1 with no invariant sections:

https://spdx.org/licenses/GFDL-1.1-no-invariants-or-later.html

So, remove the license text, replacing them by this new SPDX license.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: docs: use the new SPDX header for GFDL-1.1 on *.rst files

SPDX v3.10 gained support for GFDL-1.1 with no invariant sections:

https://spdx.org/licenses/GFDL-1.1-no-invariants-or-later.html

So, remove the license text, replacing them by this new SPDX
license.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: Make debug infrastructure more flexible

Here we introduce debug prefixes for dev_dbg groups by level of
importance - Venus{Low,Med,High,FW} Enabling the particular level
will be done by dynamic debug.

For example to enable debug messages with low level:
echo 'format "VenusLow" +p' > debugfs/dynamic_debug/control

If you want to enable all levels:
echo 'format "Venus" +p' > debugfs/dynamic_debug/control

All the features which dynamic debugging provide are preserved.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: Add debugfs interface to set firmware log level

This will be useful when debugging specific issues related to
firmware HFI interface.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: docs: Deprecate mfc frame skip control

Deprecate mfc private frame skip mode control for new
clients and use the standard one instead.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: s5p-mfc: Use standard frame skip mode control

Use the standard menu control for frame skip mode in the MFC
driver. The legacy private menu control is kept for backward
compatibility.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: venc: Add support for frame-skip mode v4l2 control

This adds support for frame-skip-mode standard v4l2 control in
encoder driver. The control is implemented based on the
combination of client selected bitrate-mode and frame-skip-mode.
Once The client selected bitrate-mode (constant or variable) and
the frame-skip-mode is not disabled we set variable framerate for
rate controller.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: v4l2-ctrl: Add frame-skip std encoder control

Adds encoders standard v4l2 control for frame-skip. The control
is a copy of a custom encoder control so that other v4l2 encoder
drivers can use it.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: venc: Add support for constant quality control

Adds implementation of V4L2_CID_MPEG_VIDEO_CONSTANT_QUALITY v4l
control when the bitrate mode is CQ.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: v4l2-ctrls: Add encoder constant quality control

When V4L2_CID_MPEG_VIDEO_BITRATE_MODE value is
V4L2_MPEG_VIDEO_BITRATE_MODE_CQ, encoder will produce
constant quality output indicated by
V4L2_CID_MPEG_VIDEO_CONSTANT_QUALITY control value.
Encoder will choose appropriate quantization parameter
and bitrate to produce requested frame quality level.

Signed-off-by: Maheshwar Ajja <majja@codeaurora.org>
Reviewed-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: Add new interface queues reinit

Presently the recovery mechanism is using two hfi functions
to destroy and create interface queues. For the purpose of
recovery we don't need to free and allocate the memory used
for interface message queues, that's why we introduce new
function which just reinit the queues.  Also this will give
to the recovery procedure one less reason to fail (if for
some reason we couldn't allocate memory).

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Fritz Koenig <frkoenig@chromium.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: Rework recovery mechanism

After power domains and clock restructuring the recovery for
sdm845 and v4 did not work properly. Fix that by reworking the
recovery function and the sequence.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Fritz Koenig <frkoenig@chromium.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: parser: Prepare parser for multiple invocations

Presently the hfi_parser has been called only once during driver
probe. To prepare the parser function to be called multiple times
from recovery we need to initialize few variables which are used
during parsing time.

Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Reviewed-by: Fritz Koenig <frkoenig@chromium.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: Fix reported frame intervals

On dragonboard-410c (apq8016) with HFI_VERSION_1XX, the reported
framerate is in unit of 1/65535 fps (for fine grained control).
So the current reported supported frame intervals is wrong (max
is 1/65535 fps), leading to encoding issues or format negotiation
failures with gstreamer.

Fix that by setting the framerate denominator to coherent value
based on the the framerate factor.

The factor is not always the same, e.g. with db820c (apq8096) HFI
reports framerate in fps unit. So only apply that for HFI_VERSION_1XX.

Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: venus: fixes for list corruption

There are few list handling issues while adding and deleting
node in the registered buf list in the driver.
1. list addition - buffer added into the list during buf_init
while not deleted during cleanup.
2. list deletion - In capture streamoff, the list was reinitialized.
As a result, if any node was present in the list, it would
lead to issue while cleaning up that node during buf_cleanup.

Corresponding call traces below:
[  165.751014] Call trace:
[  165.753541]  __list_add_valid+0x58/0x88
[  165.757532]  venus_helper_vb2_buf_init+0x74/0xa8 [venus_core]
[  165.763450]  vdec_buf_init+0x34/0xb4 [venus_dec]
[  165.768271]  __buf_prepare+0x598/0x8a0 [videobuf2_common]
[  165.773820]  vb2_core_qbuf+0xb4/0x334 [videobuf2_common]
[  165.779298]  vb2_qbuf+0x78/0xb8 [videobuf2_v4l2]
[  165.784053]  v4l2_m2m_qbuf+0x80/0xf8 [v4l2_mem2mem]
[  165.789067]  v4l2_m2m_ioctl_qbuf+0x2c/0x38 [v4l2_mem2mem]
[  165.794624]  v4l_qbuf+0x48/0x58

[ 1797.556001] Call trace:
[ 1797.558516]  __list_del_entry_valid+0x88/0x9c
[ 1797.562989]  vdec_buf_cleanup+0x54/0x228 [venus_dec]
[ 1797.568088]  __buf_prepare+0x270/0x8a0 [videobuf2_common]
[ 1797.573625]  vb2_core_qbuf+0xb4/0x338 [videobuf2_common]
[ 1797.579082]  vb2_qbuf+0x78/0xb8 [videobuf2_v4l2]
[ 1797.583830]  v4l2_m2m_qbuf+0x80/0xf8 [v4l2_mem2mem]
[ 1797.588843]  v4l2_m2m_ioctl_qbuf+0x2c/0x38 [v4l2_mem2mem]
[ 1797.594389]  v4l_qbuf+0x48/0x58

Signed-off-by: Vikash Garodia <vgarodia@codeaurora.org>
Reviewed-by: Fritz Koenig <frkoenig@chromium.org>
Signed-off-by: Stanimir Varbanov <stanimir.varbanov@linaro.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cedrus: Use H264_SCALING_MATRIX only when required

Baseline, Main and Extended profiles are specified to
not support a scaling matrix. Also, High profiles
can optionally specify a scaling matrix, using
SPS and PPS NAL units.

To meet this expectation, applications are required to
set the V4L2_CID_MPEG_VIDEO_H264_SCALING_MATRIX control
and set the V4L2_H264_PPS_FLAG_SCALING_MATRIX_PRESENT
flag only when a scaling matrix is specified for a picture.

Implement this on cedrus, which has hardware support for this
case.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: hantro: Use H264_SCALING_MATRIX only when required

Baseline, Main and Extended profiles are specified to
not support a scaling matrix. Also, High profiles
can optionally specify a scaling matrix, using
SPS and PPS NAL units.

To meet this expectation, applications are required to
set the V4L2_CID_MPEG_VIDEO_H264_SCALING_MATRIX control
and set the V4L2_H264_PPS_FLAG_SCALING_MATRIX_PRESENT
flag only when a scaling matrix is specified for a picture.

Implement this on hantro, which has hardware support for this
case.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: rkvdec: Use H264_SCALING_MATRIX only when required

Baseline, Main and Extended profiles are specified to
not support a scaling matrix. Also, High profiles
can optionally specify a scaling matrix, using
SPS and PPS NAL units.

To meet this expectation, applications are required to
set the V4L2_CID_MPEG_VIDEO_H264_SCALING_MATRIX control
and set the V4L2_H264_PPS_FLAG_SCALING_MATRIX_PRESENT
flag only when a scaling matrix is specified for a picture.

Implement this on rkvdec, which has hardware support for this
case.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: rkvdec: Drop unneeded per_request driver-specific control flag

Currently, the drivers makes no distinction between per_request
and mandatory, as both are used in the same request validate check.

The driver only cares to know if a given control is
required to be part of a request, so only one flag is needed.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cedrus: h264: Fix frame list construction

Current frame list construction algorithm assumes that decoded image
will be output into its own buffer. That is true for progressive content
but not for interlaced where each field is decoded separately into same
buffer.

Fix that by checking if capture buffer is listed in DPB. If it is, reuse
it.

Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cedrus: h264: Properly configure reference field

When interlaced H264 content is being decoded, references must indicate
which field is being referenced. Currently this was done by checking
capture buffer flags. However, that is not correct because capture
buffer may hold both fields.

Fix this by checking newly introduced flags in reference lists.

Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: rkvdec: Don't require unneeded H264_SLICE_PARAMS

Now that slice invariant parameters have been moved,
the driver no longer needs this control, so drop it.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Reviewed-by: Jonas Karlman <jonas@kwiboo.se>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: hantro: Don't require unneeded H264_SLICE_PARAMS

Now that slice invariant parameters have been moved,
the driver no longer needs this control, so drop it.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Rename and clarify PPS_FLAG_SCALING_MATRIX_PRESENT

Applications are expected to fill V4L2_CID_MPEG_VIDEO_H264_SCALING_MATRIX
if a non-flat scaling matrix applies to the picture. This is the case if
SPS scaling_matrix_present_flag or PPS pic_scaling_matrix_present_flag
are set, and should be handled by applications.

On one hand, the PPS bitstream syntax element signals the presence of a
Picture scaling matrix modifying the Sequence (SPS) scaling matrix.
On the other hand, our flag should indicate if the scaling matrix
V4L2 control is applicable to this request.

Rename the flag from PPS_FLAG_PIC_SCALING_MATRIX_PRESENT to
PPS_FLAG_SCALING_MATRIX_PRESENT, to avoid mixing this flag with
bitstream syntax element pic_scaling_matrix_present_flag,
and clarify the meaning of our flag.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Clean slice invariants syntax elements

The H.264 specification requires in section 7.4.3 "Slice header semantics",
that the following values shall be the same in all slice headers:

  pic_parameter_set_id
  frame_num
  field_pic_flag
  bottom_field_flag
  idr_pic_id
  pic_order_cnt_lsb
  delta_pic_order_cnt_bottom
  delta_pic_order_cnt[ 0 ]
  delta_pic_order_cnt[ 1 ]
  sp_for_switch_flag
  slice_group_change_cycle

These bitstream fields are part of the slice header, and therefore
passed redundantly on each slice. The purpose of the redundancy
is to make the codec fault-tolerant in network scenarios.

This is of course not needed to be reflected in the V4L2 controls,
given the bitstream has already been parsed by applications.
Therefore, move the redundant fields to the per-frame decode
parameters control (DECODE_PARAMS).

Field 'pic_parameter_set_id' is simply removed in this case,
because the PPS control must currently contain the active PPS.

Syntax elements dec_ref_pic_marking() and those related
to pic order count, remain invariant as well, and therefore,
the fields dec_ref_pic_marking_bit_size and pic_order_cnt_bit_size
are also common to all slices.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Clarify SLICE_BASED mode

Currently, the SLICE_BASED and FRAME_BASED modes documentation
is misleading and not matching the intended use-cases.

Drop non-required fields SLICE_PARAMS 'start_byte_offset' and
DECODE_PARAMS 'num_slices' and clarify the decoding modes in the
documentation.

On SLICE_BASED mode, a single slice is expected per OUTPUT buffer,
and therefore 'start_byte_offset' is not needed (since the offset
to the slice is the start of the buffer).

This mode requires the use of CAPTURE buffer holding, and so
the number of slices shall not be required.

On FRAME_BASED mode, the devices are expected to take care of slice
parsing. Neither SLICE_PARAMS are required (and shouldn't be
exposed by frame-based drivers), nor the number of slices.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Drop SLICE_PARAMS 'size' field

The SLICE_PARAMS control is intended for slice-based
devices. In this mode, the OUTPUT buffer contains
a single slice, and so the buffer's plane payload size
can be used to query the slice size.

To reduce the API surface drop the size from the
SLICE_PARAMS control.

A follow-up change will remove other members in SLICE_PARAMS
so we don't need to add padding fields here.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Increase size of DPB entry pic_num

DPB entry PicNum maximum value is 2*MaxFrameNum for interlaced
content (field_pic_flag=1).

As specified, MaxFrameNum is 2^(log2_max_frame_num_minus4 + 4)
and log2_max_frame_num_minus4 is in the range of 0 to 12,
which means pic_num should be a 32-bit field.

The v4l2_h264_dpb_entry struct needs to be padded to avoid a hole,
which might be also useful to allow future uAPI extensions.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Clean DPB entry interface

As discussed recently, the current interface for the
Decoded Picture Buffer is not enough to properly
support field coding.

This commit introduces enough semantics to support
frame and field coding, and to signal how DPB entries
are "used for reference".

Reserved fields will be added by a follow-up commit.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Increase size of 'first_mb_in_slice' field

Slice header syntax element 'first_mb_in_slice' can point
to the last macroblock, currently the field can only reference
65536 macroblocks which is insufficient for 8K videos.

Although unlikely, a 8192x4320 video (where macroblocks are 16x16),
would contain 138240 macroblocks on a frame.

As per the H264 specification, 'first_mb_in_slice' can be up to
PicSizeInMbs - 1, so increase the size of the field to 32-bits.

Note that v4l2_ctrl_h264_slice_params struct will be modified
in a follow-up commit, and so we defer its 64-bit padding.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Clarify pic_order_cnt_bit_size field

Since pic_order_cnt_bit_size is not a syntax element itself, explicitly
state that it is the total size in bits of the pic_order_cnt_lsb,
delta_pic_order_cnt_bottom, delta_pic_order_cnt[0], and
delta_pic_order_cnt[1] syntax elements contained in the slice.

[Ezequiel: rebase]

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Reviewed-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Split prediction weight parameters

The prediction weight parameters are only required under
certain conditions, which depend on slice header parameters.

As specified in section 7.3.3 Slice header syntax, the prediction
weight table is present if:

((weighted_pred_flag && (slice_type == P || slice_type == SP)) || \
(weighted_bipred_idc == 1 && slice_type == B))

Given its size, it makes sense to move this table to its control,
so applications can avoid passing it if the slice doesn't specify it.

Before this change struct v4l2_ctrl_h264_slice_params was 960 bytes.
With this change, it's 188 bytes and struct v4l2_ctrl_h264_pred_weight
is 772 bytes.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Further clarify scaling lists order

Commit 0b0393d59eb4a ("media: uapi: h264: clarify
expected scaling_list_4x4/8x8 order") improved the
documentation on H264 scaling lists order.

This commit improves the documentation by clarifying
that the lists themselves are expected in raster scan order.

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: uapi: h264: Update reference lists

When dealing with interlaced frames, reference lists must tell if
each particular reference is meant for top or bottom field. This info
is currently not provided at all in the H264 related controls.

Change reference lists to hold a structure, which specifies
an index into the DPB array and the field/frame specification
for the picture.

Currently the only user of these lists is Cedrus which is just compile
fixed here. Actual usage of will come in a following commit.

Signed-off-by: Jernej Skrabec <jernej.skrabec@siol.net>
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Tested-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: ov5675: correct the maximum exposure value

The unit of exposure value is different from other OmniVision sensors,
driver will divide by 2 before set register, the exposure range exposed
by v4l2 ctrl to user should be same as others, so the calculation for
the maximum exposure value in current driver need be fixed.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: Documentation: media: Document how to write camera sensor drivers

While we have had some example drivers, there has been up to date no
formal documentation on how camera sensor drivers should be written; what
are the practices, why, and where they apply.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: ov7740: use SCCB regmap

Make use of SCCB APIs for regmap operations.
Remove i2c_check_functionality as devm_regmap_init_sccb
does it for us.

Signed-off-by: Raag Jadav <raagjadav@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: imx274: fix frame interval handling

1) the numerator and/or denominator might be 0, in that case
   fall back to the default frame interval. This is per the spec
   and this caused a v4l2-compliance failure.

2) the updated frame interval wasn't returned in the s_frame_interval
   subdev op.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Reviewed-by: Luca Ceresoli <luca@lucaceresoli.net>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: m5mols: Check function pointer in m5mols_sensor_power

clang static analysis reports this error

m5mols_core.c:767:4: warning: Called function pointer
  is null (null dereference) [core.CallAndMessage]
    info->set_power(&client->dev, 0);
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

In other places, the set_power ptr is checked.
So add a check.

Fixes: bc125106f8af ("[media] Add support for M-5MOLS 8 Mega Pixel camera ISP")
Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: i2c: fix error check on max9286_read call

Currently the error return from the call to max9286_read is masked
with 0xf0 so the following check for a negative error return is
never true.  Fix this by checking for an error first, then masking
the return value for subsequent conflink_mask checking.

Addresses-Coverity: ("Logically dead code")

Fixes: 66d8c9d2422d ("media: i2c: Add MAX9286 driver")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: omap: Replace HTTP links with HTTPS ones

Rationale:
Reduces attack surface on kernel devs opening the links for MITM
as HTTPS traffic is much harder to manipulate.

Deterministic algorithm:
For each file:
  If not .svg:
    For each line:
      If doesn't contain `\bxmlns\b`:
  If neither `\bgnu\.org/license`, nor `\bmozilla\.org/MPL\b`:
            If both the HTTP and HTTPS versions
            return 200 OK and serve the same content:
              Replace HTTP with HTTPS.

Signed-off-by: Alexander A. Klimov <grandmaster@al2klimov.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: ov5640: Correct Bit Div register in clock tree diagram

Although the code is correct and doing the right thing, the clock diagram
showed the wrong register for the bit divider, which had me doubting the
understanding of the tree. Fix this to avoid doubts in the future.

Signed-off-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Fixes: aa2882481cada ("media: ov5640: Adjust the clock based on the expected rate")
Acked-by: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: ipu3.rst: Format media-ctl and yavta commands as code blocks

Fix improper line breaks and format all example yavta and media-ctl
commands as code blocks to improve readability.

Signed-off-by: Jordan Hand <jorhand@linux.microsoft.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: MAINTAINERS: ov5647: Replace maintainer

Since the current maintainer email address bounces back, replace
the entry and make myself and Dave Stevenson maintainers of the driver.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Acked-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: dt-bindings: media: ov5647: Document clock-noncontinuous

Document the optional clock-noncontinuous endpoint property that
allows enabling MIPI CSI-2 non-continuous clock operations.

Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: dt-bindings: media: ov5647: Document pwdn-gpios

Document in dt-schema bindings for the ov5647 sensor the optional
'pwdn-gpios' property.

Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: dt-bindings: media: ov5647: Convert to json-schema

Convert the ov5647 image sensor bindings to DT schema and add
the file entry to MAINTAINERS.

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: intel-ipu3: Fix code style issue

This fixes a multiline comment style warning as found by checkpatch.pl.

Signed-off-by: Rahul Gottipati <rahul.blr97@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: MAINTAINERS: Fix email typo and correct name of Tianshu

Fix the typo in email address of Tianshu Qiu and correct the name.

Reported-by: Bjorn Helgaas <helgaas@kernel.org>
Signed-off-by: Tianshu Qiu <tian.shu.qiu@intel.com>
Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: i2c: ov2740: get OTP data ready before nvmem registration

The OTP data was not ready after registered as nvmem device, it is
risky as the nvmem read may happen once the device exists, this patch
get the OTP data ready before registering the nvmem device. OTP data
missing should not break the normal camera sensor probe, so use a
warning instead of an error message.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Qingwu Zhang <qingwu.zhang@intel.com>
Suggested-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: radio: si476x: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: sti: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: mtk-vpu: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: coda: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: cec: no need to check return value of debugfs_create functions

When calling debugfs functions, there is no need to ever check the
return value.  The function can work or not, but the code logic should
never do something different based on this.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: dt-bindings: media: Correct samsung-fimc parallel port numbering

The parallel port nodes should be numbered 1 and 2, not 0 and 1
for A and B respectively.  The driver has always implemented 1
and 2 and the in-tree Goni DTS uses 1 as port A as well.  Update
the documentation to match this behaviour.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Handle duplicate calls to vidioc_streamoff

vidioc_streamoff can be called multiple times from userspace, but we
should only call media_pipeline_stop when we're actually setup.

This became more noticeable after commit 2a2599c66368 ("[media] media:
entity: Catch unbalanced media_pipeline_stop calls") was merged as it
added a WARN for unbalanced calls to media_pipeline_stop.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Remove unused struct member input_index

This is no longer used since the conversion to DT

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Remove inh_sensor_ctrls

This is a no-op as it is never set and is a remnant from non-DT days
that can be safely removed.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Add support for multiple sensors on one port

On some devices, there may be multiple camera sensors attached
to the same port.  Make sure we probe all of them, not just the
first one.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Use global num_sensors rather than local index

Instead of keeping a local copy of how many sensors we've probed
(which may not even properly represent the number of sensors
probed if we have a port without a sensor), use the global
num_sensors counter that has the actual number used.

This will also make it easier to add support for multiple sensors
being connected to the same port.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Properly set JPEG options for parallel ports

Commit ee7160e57c98 ("[media] s5p-fimc: Add support for JPEG capture")
added support for JPEG capture, but missed setting a register when the
parallel port was used rather than the CSIS device.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Correct missing entity function initialization

Commit bae4500399c4 ("[media] exynos4-is: Add missing entity function
initialization") tried to suppress the warnings such as

s5p-fimc-md camera: Entity type for entity FIMC.0 was not initialized!

However, this missed setting for the subdev.  Set it now to avoid the
ugly warnings on boot.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Fix nullptr when no CSIS device present

Not all devices use the CSIS device, some may use the FIMC directly in
which case the CSIS device isn't registered.  This leads to a nullptr
exception when starting the stream as the CSIS device is always
referenced.  Instead, if getting the CSIS device fails, try getting the
FIMC directly to check if we are using the subdev API

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Request syscon only if ISP writeback is present

On FIMC variants which don't have writeback channel, there is no need to
access system registers. This patch makes the driver request sysreg
regmap conditionally depending on whether writeback is supported.

Signed-off-by: Tomasz Figa <tomasz.figa@gmail.com>
Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: exynos4-is: Remove static driver data for S5PV210 FIMC variants

The S5PV210 platform only supports device tree based booting
where the FIMC variant data is parsed directly from
the device tree, hence the now unused static data can be removed.

Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Reviewed-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: hantro: postproc: Fix motion vector space allocation

When the post-processor is enabled, the driver allocates
"shadow buffers" which are used for the decoder core,
and exposes the post-processed buffers to userspace.

For this reason, extra motion vector space has to
be allocated on the shadow buffers, which the driver
wasn't doing. Fix it.

This fix should address artifacts on high profile bitstreams.

Fixes: 8c2d66b036c77 ("media: hantro: Support color conversion via post-processing")
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>

media: hantro: h264: Get the correct fallback reference buffer

If the bitstream and the application are incorrectly configuring
the reference pictures, the hardware will need to fallback
to using some other reference picture.

When the post-processor is enabled, the fallback buffer
should be a shadow buffer (postproc.dec_q), and not a
CAPTURE queue buffer, since the latter is post-processed
and not really the output of the decoder core.

Fixes: 8c2d66b036c77 ("media: hantro: Support color conversion via post-processing")
Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>