media: intel/ipu6: add Kconfig and Makefile

Add Kconfig and Makefile for the IPU6 driver.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Andreas Helbech Kleist <andreaskleist@gmail.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add the main input system driver

The main input system driver does basic ISYS hardware setup, IRQ handling
and sets up device nodes.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: input system video nodes and buffer queues

Register V4L2 video device and setup the VB2 queues to support video
capture. Video streaming callback will trigger the input system driver to
construct a input system stream configuration for firmware based on data
type and stream ID and then queue buffers to firmware to do capture.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Co-developed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add the CSI2 DPHY implementation

IPU6 CSI-2 D-PHY hardware varies on different platforms, current IPU6 has
three D-PHY hardware instances which are used on Tigerlake, Alder lake,
Meteor lake and Jasper lake. MCD D-PHY is included in Tigerlake and Alder
lake, DWC D-PHY is included in Meteor lake.

Each PHY has its own register interface, the input system driver calls the
appropriate D-PHY callbacks for the hardware set in isys_probe().

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add IPU6 CSI2 receiver v4l2 sub-device

Input system CSI2 receiver is exposed as a v4l2 sub-device. Each CSI2
sub-device represent one single CSI2 hardware port which be linked with
external sub-device such camera sensor by linked with ISYS CSI2's sink
pad. The CSI2 source pad is linked to the sink pad of video capture device.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: input system ABI between firmware and driver

Implement the input system firmware ABIs in the ISYS driver, including
stream configuration, control command, capture request and response.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Co-developed-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add syscom interfaces between firmware and driver

Syscom is an inter-process(or) communication mechanism between an IPU and
host. Syscom uses message queues for message exchange between IPU and
host. Each message queue has its consumer and producer, host queue
messages to firmware as the producer and then firmware to dequeue the
messages as consumer and vice versa. IPU and host use shared registers or
memory to reside the read and write indices which are updated by consumer
and producer.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add IPU6 DMA mapping API and MMU table

The Intel IPU6 has internal microcontrollers (scalar processor, SP) which
are used to execute the firmware. The SPs can access IPU internal memory
and system DRAM mapped to its an internal 32-bit virtual address space.

This patch adds a driver for the IPU MMU and a DMA mapping implementation
using the internal MMU. The system IOMMU may be used besides the IPU MMU.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: CPD parsing for get firmware components

For IPU6, firmware is generated and released as signed Code Partition
Directory (CPD) format file which is aligned with the SPI flash code
partition definition. The CPD format includes CPD header, manifest,
metadata and module data. The driver parses them according to the CPD
layout to acquire each component.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add IPU6 buttress interface driver

The IPU6 buttress is the interface between IPU device (input system and
processing system) with rest of the SoC. It contains overall IPU hardware
control registers, these control registers are used as the interfaces with
the Intel Converged Security Engine and Punit to do firmware
authentication and power management.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add IPU auxiliary devices

Even though the IPU input system and processing system are in a single PCI
device, each system has its own power sequence. Powering up the processing
system depends on first powering up the input system.

Besides, both the input and processing systems have their own MMU hardware
for IPU DMA address mapping.

Register the IS/PS devices on auxiliary bus and attach power domain to
implement the power sequence dependency.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: intel/ipu6: add Intel IPU6 PCI device driver

Intel Image Processing Unit 6th Gen includes input and processing systems
but the hardware presents itself as a single PCI device in system.

The IPU6 PCI device driver basically does PCI configurations and loads the
firmware binary, initialises IPU virtual bus and sets up platform specific
variants to support multiple IPU6 devices in single device driver.

Signed-off-by: Bingbu Cao <bingbu.cao@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: ivsc: csi: Use IPU bridge

Use IPU bridge to instantiate software nodes for IPU6 related devices. If
the IPU6 device is probed before the MEI CSI device is created, neither
will probe unless there are further devices probed successfully which
triggers re-probe of both IPU6 and MEI CSI.

This patch ensures the necessary software nodes are in place when the MEI
CSI driver is probed.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: ipu6: Add PCI device table header

Add a header for the PCI device table of all IPU6 devices. This will be
used by the IPU bridge as well so place the table in a header.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: bcm2835-unicam: Add support for CCP2/CSI2 camera interface

Add a driver for the Unicam camera receiver block on BCM283x processors.
It is represented as two video device nodes: unicam-image and
unicam-embedded which are connected to an internal subdev (named
unicam-subdev) in order to manage streams routing.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Co-developed-by: Naushir Patuck <naush@raspberrypi.com>
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Co-developed-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Co-developed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
[Sakari Ailus: Squash fixes by Laurent.]
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

dt-bindings: media: Add bindings for bcm2835-unicam

Introduce the dt-bindings documentation for bcm2835 CCP2/CSI2 Unicam
camera interface.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Co-developed-by: Naushir Patuck <naush@raspberrypi.com>
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Co-developed-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: Add V4L2-PIX-FMT-Y14P format

This is a packed grey-scale image format with a depth of 14 bits per
pixel. Every four consecutive samples are packed into seven bytes. Each
of the first four bytes contain the eight high order bits of the pixels,
and the three following bytes contains the six least significants bits
of each pixel, in the same order.

As the other formats only needed 5 bytes before, append two bytes in the
documentation array.

Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: Add V4L2-PIX-FMT-Y12P format

This is a packed grey-scale image format with a depth of 12 bits per
pixel. Two consecutive pixels are packed into 3 bytes. The first 2 bytes
contain the 8 high order bits of the pixels, and the 3rd byte contains
the 4 least significants bits of each pixel, in the same order.

Add the entry in userspace API, and document it.

Signed-off-by: Jean-Michel Hautbois <jeanmichel.hautbois@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: subdev: Add trivial set_routing support

Add trivial S_ROUTING IOCTL support for drivers where routing is static.
Essentially this means returning the same information G_ROUTING call would
have done.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Co-developed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Julien Massot <julien.massot@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: subdev: Return routes set using S_ROUTING

Return the routes set using S_ROUTING back to the user. Also reflect this
in documentation.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: subdev: Add len_routes field to struct v4l2_subdev_routing

The len_routes field is used to tell the size of the routes array in
struct v4l2_subdev_routing. This way the number of routes returned from
S_ROUTING IOCTL may be larger than the number of routes provided, in case
there are more routes returned by the driver.

Note that this uAPI is still disabled in the code, so this change can
safely be done. Anyone who manually patched the code to enable this uAPI
must update their code.

The patch also increases the number of reserved fields in struct
v4l2_subdev_routing.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: subdev: Copy argument back to user also for S_ROUTING

As the user needs to know what went wrong for S_ROUTING, copy array
arguments back to the user.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Julien Massot <julien.massot@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: subdev: Add a function to lock two sub-device states, use it

Add two new functions, v4l2_subdev_lock_states() and
v4l2_subdev_unclock_states(), to acquire and release the state of two
sub-devices. They differ from calling v4l2_subdev_{un,}lock_state() so
that if the two states share the same lock, the lock is acquired only
once.

Also use the new functions in v4l2_subdev_link_validate().

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Julien Massot <julien.massot@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: Documentation: Document S_ROUTING behaviour

Document S_ROUTING behaviour for different devices.

Generally in devices that produce streams the streams are static and some
can be enabled and disabled, whereas in devices that just transport them
or write them to memory, more configurability is allowed. Document this.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Julien Massot <julien.massot@collabora.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: Documentation: Additional streams generally don't harm capture

Having extra streams on the source end of the link that cannot be captured
by the sink sub-device generally are not an issue, at least not on CSI-2
bus. Still document that there may be hardware-specific limitations. For
example on parallel bus this might not work on all cases.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: Set line based metadata flag in V4L2 core

Set (and unset) the V4L2_FMT_FLAG_META_LINE_BASED flag in struct
v4l2_fmtdesc based on the format after returning the driver callback for
enumerating formats. This way the drivers don't need to care about the
flag.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l: Support line-based metadata capture

Many camera sensors, among other devices, transmit embedded data and image
data for each CSI-2 frame. This embedded data typically contains register
configuration of the sensor that has been used to capture the image data
of the same frame.

The embedded data is received by the CSI-2 receiver and has the same
properties as the image data, including that it is line based: it has
width, height and bytesperline (stride).

Add these fields to struct v4l2_meta_format and document them.

Also add V4L2_FMT_FLAG_META_LINE_BASED to tell a given format is
line-based i.e. these fields of struct v4l2_meta_format are valid for it.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: uapi: v4l: Add generic 8-bit metadata format definitions

Generic 8-bit metadata formats define the in-memory data layout but not
the format of the data itself. The reasoning for having such formats is to
allow CSI-2 receiver drivers to receive and DMA drivers to write the data
to memory without knowing a large number of device-specific formats.

These formats may be used only in conjunction with a Media Controller
pipeline where the internal pad of the source sub-device defines the
specific format of the data (using an mbus code).

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: uapi: Document which mbus format fields are valid for metadata

Now that metadata mbus formats have been added, it is necessary to define
which fields in struct v4l2_mbus_format are applicable to them (not many).

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: uapi: Add generic serial metadata mbus formats

Add generic serial metadata mbus formats. These formats describe data
width and packing but not the content itself. The reason for specifying
such formats is that the formats as such are fairly device specific but
they are still handled by CSI-2 receiver drivers that should not be aware
of device specific formats. What makes generic metadata formats possible
is that these formats are parsed by software only, after capturing the
data to system memory.

Also add a definition for "Data Unit" to cover what is essentially a pixel
but is not image data.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: Documentation: Add "stream" into glossary

Add term "stream" to the glossary of the Media subsystem documentation.

Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: v4l2-subdev: Clearly document that the crop API won't be extended

The V4L2 subdev crop API has been marked as obsolete, deprecated by the
selection API. Despite this, it has recently been extended with streams
support. In hindsight this was a mistake. Make sure it doesn't happen
again by clearly documenting that no new extensions will be accepted.

Suggested-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Acked-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: si2165: Remove redundant NULL check before release_firmware() call

release_firmware() checks for NULL pointers internally so checking
before calling it is redundant.

Link: https://lore.kernel.org/linux-media/20220606014433.290667-1-chi.minghao@zte.com.cn
Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: Minghao Chi <chi.minghao@zte.com.cn>
Acked-by: Matthias Schwarzott <zzam@gentoo.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: flexcop-usb: fix sanity check of bNumEndpoints

Commit d725d20e81c2 ("media: flexcop-usb: sanity checking of endpoint type
") adds a sanity check for endpoint[1], but fails to modify the sanity
check of bNumEndpoints.

Fix this by modifying the sanity check of bNumEndpoints to 2.

Link: https://lore.kernel.org/linux-media/20220602055027.849014-1-dzm91@hust.edu.cn
Fixes: d725d20e81c2 ("media: flexcop-usb: sanity checking of endpoint type")
Signed-off-by: Dongliang Mu <mudongliangabcd@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: flexcop: allow for modern speeds

High speed is no longer the ultimate in speed.

Link: https://lore.kernel.org/linux-media/20220517131109.28371-2-oneukum@suse.com
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: flexcop: unneeded ATOMIC

No need for GFP_ATOMIC during probe()

Link: https://lore.kernel.org/linux-media/20220517131109.28371-1-oneukum@suse.com
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Reviewed-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: as102: avoid GFP_ATOMIC

No need for GFP_ATOMIC during probe()

Link: https://lore.kernel.org/linux-media/20220517111049.25611-1-oneukum@suse.com
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: imx214: Fix the error handling in imx214_probe()

The driver should disable regulators when fails to probe.

Link: https://lore.kernel.org/linux-media/20220510114852.1719018-1-zheyuma97@gmail.com
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: lgdt3306a: Add a check against null-pointer-def

The driver should check whether the client provides the platform_data.

The following log reveals it:

[   29.610324] BUG: KASAN: null-ptr-deref in kmemdup+0x30/0x40
[   29.610730] Read of size 40 at addr 0000000000000000 by task bash/414
[   29.612820] Call Trace:
[   29.613030]  <TASK>
[   29.613201]  dump_stack_lvl+0x56/0x6f
[   29.613496]  ? kmemdup+0x30/0x40
[   29.613754]  print_report.cold+0x494/0x6b7
[   29.614082]  ? kmemdup+0x30/0x40
[   29.614340]  kasan_report+0x8a/0x190
[   29.614628]  ? kmemdup+0x30/0x40
[   29.614888]  kasan_check_range+0x14d/0x1d0
[   29.615213]  memcpy+0x20/0x60
[   29.615454]  kmemdup+0x30/0x40
[   29.615700]  lgdt3306a_probe+0x52/0x310
[   29.616339]  i2c_device_probe+0x951/0xa90

Link: https://lore.kernel.org/linux-media/20220405095018.3993578-1-zheyuma97@gmail.com
Signed-off-by: Zheyu Ma <zheyuma97@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: dw2102: fix coding style issues

Use checkpatch --strict --in-place to cleanup most style issues.
Then, do some manual work to fix most of the remaining issues.

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

media: dw2102: Fix minor issues

Found by checkpatch.

Link: https://lore.kernel.org/linux-media/20220116112238.74171-4-micha@freedict.org
Signed-off-by: Michael Bunk <micha@freedict.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: dw2102: Dont't list TT Cinergy S2 R4 twice

"modinfo dvb-usb-dw2102|grep 0105" returns two lines, because the module
reports this usb alias twice.

Remove one of them.

Signed-off-by: Michael Bunk <micha@freedict.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: dw2102: Don't translate i2c read into write

The code ignored the I2C_M_RD flag on I2C messages.  Instead it assumed
an i2c transaction with a single message must be a write operation and a
transaction with two messages would be a read operation.

Though this works for the driver code, it leads to problems once the i2c
device is exposed to code not knowing this convention.  For example,
I did "insmod i2c-dev" and issued read requests from userspace, which
were translated into write requests and destroyed the EEPROM of my
device.

So, just check and respect the I2C_M_READ flag, which indicates a read
when set on a message.  If it is absent, it is a write message.

Incidentally, changing from the case statement to a while loop allows
the code to lift the limitation to two i2c messages per transaction.

There are 4 more *_i2c_transfer functions affected by the same behaviour
and limitation that should be fixed in the same way.

Link: https://lore.kernel.org/linux-media/20220116112238.74171-2-micha@freedict.org
Signed-off-by: Michael Bunk <micha@freedict.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: m88ds3103: Fix lock led support

The new m88ds3101b demodulator version requires the highest bit of
register 0x11 be cleared after tuning to a new frequency to make the
lock output pin go high when a signal is found.  This pin is normally
connected to an indicator led.

Link: https://lore.kernel.org/linux-media/20220116112238.74171-1-micha@freedict.org
Signed-off-by: Michael Bunk <micha@freedict.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: i2c: max9271: Add header include guards to max9271.h

Protect against multiple inclusions of the max9271.h header with include
guards.

Link: https://lore.kernel.org/linux-media/20220102224803.27463-1-laurent.pinchart+renesas@ideasonboard.com
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: i2c: rdacm20: Fix indentation in comment

Fix a small indentation issue in a comment block.

Link: https://lore.kernel.org/linux-media/20220101173540.9090-1-laurent.pinchart+renesas@ideasonboard.com
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: fix possible unbalanced PM counter

It is possible that mtk_vcodec_enc_pw_on fails, and in that scenario
the PM counter is not incremented, and subsequent call to
mtk_vcodec_enc_pw_off decrements the counter, leading to a PM imbalance.
Fix by bailing out of venc_if_encode in the case when mtk_vcodec_enc_pw_on
fails.

Fixes: 4e855a6efa54 ("[media] vcodec: mediatek: Add Mediatek V4L2 Video Encoder Driver")
Signed-off-by: Eugen Hristev <eugen.hristev@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Signed-off-by: Sebastian Fricke <sebastian.fricke@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>

media: mediatek: vcodec: Only free buffer VA that is not NULL

In the MediaTek vcodec driver, while mtk_vcodec_mem_free() is mostly
called only when the buffer to free exists, there are some instances
that didn't do the check and triggered warnings in practice.

We believe those checks were forgotten unintentionally. Add the checks
back to fix the warnings.

Signed-off-by: Fei Shao <fshao@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Signed-off-by: Sebastian Fricke <sebastian.fricke@collabora.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: Update mtk_vcodec_mem_free() error messages

In mtk_vcodec_mem_free(), there are two cases where a NULL VA is passed:
- mem->size == 0: we are called to free no memory. This may happen when
  we call mtk_vcodec_mem_free() twice or the memory has never been
  allocated.
- mem->size > 0: we are called to free memory but without VA. This means
  that we failed to free the memory for real.

Both cases are not expected to happen, and we want to have clearer error
messages to describe which one we just encountered.
Update the error messages to include more information for that purpose.

Signed-off-by: Fei Shao <fshao@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Signed-off-by: Sebastian Fricke <sebastian.fricke@collabora.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: Drop unnecessary variable

In mtk_vcodec_mem_alloc() and mtk_vcodec_mem_free(), the value of
mem->size is not expected to change before and when using the DMA APIs
and debug print, so there's no point in keeping local copies of it.

Drop the local variable "size" in the mentioned functions, and update
printk format identifiers accordingly.

This makes the code slightly more visually consistent, and retrieve a
small amount of memory that is used for no real purpose.

Signed-off-by: Fei Shao <fshao@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Signed-off-by: Sebastian Fricke <sebastian.fricke@collabora.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: mediatek: vcodec: Replace dev_name in error string

mtk_v4l2_err() already uses dev_err(), so don't print the device name
again. Print the function name instead.

Signed-off-by: Fei Shao <fshao@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Andrzej Pietrasiewicz <andrzej.p@collabora.com>
Signed-off-by: Sebastian Fricke <sebastian.fricke@collabora.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: v4l: async: Set owner for async sub-devices

Set the owner field of the async sub-devices by making
v4l2_async_register_subdev() a macro and obtaining THIS_MODULE that way.

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

media: v4l: Set sub-device's owner field to the caller's module

Set a sub-device's owner field to the caller's module, provided as an
argument to the function. v4l2_device_register_subdev() becomes a macro
passing THIS_MODULE to the __v4l2_device_register_subdev() function.

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

media: i2c: imx219: Use dev_err_probe on probe

Drop dev_err() and use the dev_err_probe() helper on probe path.

No functional changes intended.

Signed-off-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: i2c: ov4689: Set timing registers programmatically

Set timing-related and BLC anchor registers via cci calls instead of
hardcoding them in the register table. This prepares the driver for
implementation of configurable analogue crop and binning. No
functional change intended.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Move pixel array size out of struct ov4689_mode

Pixel array dimensions and default crop size do not belong to the
ov4689_mode structure, since they are mode independent. Make them
defines instead.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: i2c: ov4689: Implement manual color balance controls

The OV4689 sensor has separate red and blue gain settings (up to 4x).
Implement appropriate controls in the driver. Default gain values
are not modified.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Implement digital gain control

The OV4689 sensor supports digital gain up to 16x. Implement
corresponding control in the driver. Default digital gain value is not
modified by this patch.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Implement vflip/hflip controls

The OV4689 sensor supports horizontal and vertical flipping. Add
appropriate controls to the driver. Toggling both array flip and
digital flip bits allows to achieve flipping while maintaining output
Bayer order. Note that the default value of hflip control corresponds
to both bits set, as it was before this patch.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: i2c: ov4689: Make horizontal blanking configurable

Make horizontal blanking configurable. To do so, set HTS register
according to the requested horizontal blanking in ov4689_set_ctrl
instead of the register table. Default HTS value is not changed by
this patch. Minimal HTS value is found experimentally and corresponds
to 90 fps framerate at minimum vertical blanking. Real HTS value is
the register value multiplied by 4.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Remove max_fps field from struct ov4689_mode

max_fps field of struct ov4689_mode is unused in this driver, so
remove it.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Use runtime PM autosuspend

Use runtime PM autosuspend to avoid powering off the sensor during
fast stop-reconfigure-restart cycles.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Enable runtime PM before registering sub-device

As the sensor may be accessible right after its async sub-device is
registered, enable runtime PM before doing so.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Use sub-device active state

Use sub-device active state. Employ control handler lock to
synchronize access to the active state and s_stream.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Refactor ov4689_set_ctrl

Introduce local variable for regmap within the ov4689_set_ctrl
function. This adjustment eliminates repetition within the function.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: i2c: ov4689: Remove i2c_client from ov4689 struct

The 'client' field within the 'ov4689' structure is solely used to
access its 'dev' member. This commit removes the 'client' field and
directly stores a pointer to the 'struct device'.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: CCI conversion

Convert the i2c register accesses to utilize the CCI helper library
rather than relying on driver-specific functions. Also, set analogue
gain in a single 16-bit write instead of two 8-bit writes.

[Sakari Ailus: Initialise sensor_gain as 0.]

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Fix typo in a comment

Fix a spelling error in a comment.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Sort register definitions by address

Put register defininitions in the order of increasing register
address.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
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@kernel.org>

media: i2c: ov4689: Clean up and annotate the register table

Many values in the register table are actually power-on
defaults. Remove those and also unused HDR exposures and gains.
Annotate the remaining values using the publicly available datasheet
to facilitate further development. No functional change intended.

Signed-off-by: Mikhail Rudenko <mike.rudenko@gmail.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Fix sh_css_config_input_network_2400() coding style

Fix the if (stream->config.mode == IA_CSS_INPUT_MODE_PRBS) {} block
coding style. Specifically fix the weird declaration of variables split
over multiple lines.

While at it also split the multi-line width calculation at a more
logical place.

Reported-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove setting of f->fmt.pix.priv from atomisp_set_fmt()

This value is not used by the driver, only returned to userspace and
userspace should not rely on / use the priv fields. Drop setting this.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Cleanup atomisp_isr_thread() spinlock handling

Refactor the code a tiny bit to avoid the need to have 2 different
paths with spin_unlock_irqrestore() in there.

While at it also remove the non helpful dev_dbg() message.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove empty s_stream() op from CSI subdev

Remove the empty csi2_set_stream() callback and remove the now
empty csi2_video_ops struct.

While at it also remove the empty csi2_core_ops struct.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove empty s_power() op from ISP subdev

Remove the empty isp_subdev_set_power() callback.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove clearing of ISP crop / compose rectangles on file release

There is no need to clear the ISP sink-pad crop and source-pad
compose rectangles  on release of the /dev/video# file.

Both will be reset by atomisp_set_fmt() which is guaranteed to
get called by atomisp_queue_setup() (if not called already)
before another stream starts and thus before these selections
are used again.

Also the file injection support the comment talks about has long
been dropped from the driver.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove struct atomisp_platform_data

struct atomisp_platform_data only has a single subdevs member. Remove
this struct and replace atomisp_get_platform_data() with
atomisp_platform_get_subdevs().

Note atomisp_get_platform_data() and now atomisp_platform_get_subdevs()
always succeed, so while at it also remove unnecessary error checking.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove gmin_platform VCM code.

VCM support has already been removed from the main atomisp driver
in commit 7473c97dd8d0 ("media: atomisp: Remove custom VCM handling")
remove the left-over dead VCM handling from the gmin platform code.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Drop intel_v4l2_subdev_type

intel_v4l2_subdev_type / atomisp_input_subdev.type now always is
RAW_CAMERA, drop it.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Drop unused frame_status tracking

After removing the flash support frame_status now always is
ATOMISP_FRAME_STATUS_OK aka 0. Drop frame_status tracking.

This also means one less atomisp custom use for the vb2_buffer->reserved*
fields (all custom use of these fields should be removed).

While at it also remove the no longer used vf_frame pointer from
struct atomisp_sub_device.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Drop custom flash support

Remove the custom flash handling from the atomisp driver. There are
only few Bay Trail / Cherry Trail devices with flash and if those will
ever get supported then this should be done through the new standard
include/linux/led-class-flash.h APIs instead of with atomisp specific
custom APIs.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Drop the atomisp custom lm3554 flash driver

Drop the atomisp custom lm3554 flash driver, support for flash drivers
using the custom atomisp flash interface is going to be removed from
the atomisp driver.

If flash supports for the lm3554 turns out to be necessary on some
atomisp devices in the future then this should be re-implemented
using the new include/linux/led-class-flash.h APIs.

The atomisp specific lm3554 driver could still be usable as a reference
for how to use / driver the lm3554 for a future re-implementation
using standardized APIs, but for that it can be retrieved from
the git history.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove input_port_ID_t

Change the single, unexpected user of CSI_PORT0_ID

for (port = CSI_PORT0_ID; port < N_CSI_PORTS; port++)

to

for (port = 0; port < N_CSI_PORTS; port++) {

matching all the other for-loops iterating over the ports in
the same file.

And remove the now fully unused input_port_ID_t enum type.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove test pattern generator (TPG) support

The TPG support registers a v4l2-subdev for this, but this is not part of
the media-controller graph in any way. After manually selecting the TPG
as input using the s_input ioctl it does not work.

Several supported sensors have their own working TPG and using the sensor's
TPG means that the same data-flow is used as with actual sensors rather
then the quite different data-flow from the ISP's builtin TPG.

Remove the broken TPG support, when a test-pattern is needed for testing
a sensor's TPG can be used. Using a sensor's TPG is actually better for
testing since then the actual normal data-flow is being tested.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: ov2722: Remove power on sensor from set_fmt() callback

The atomisp driver now properly ensures s_power(1) is called before calling
the set_fmt() callback, so this workaround is no longer necessary.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Drop atomisp_pipe_check() from atomisp_link_setup()

The media-controller core (__media_entity_setup_link()) already checks
that the pads of the link are not streaming before calling the setup_link()
pad-op so calling atomisp_pipe_check() is not necessary;

and taking isp->mutex inside the setup_link() pad-op leads to a possible
ABBA deadlock vs the media-device graph_mutex which in the case of
the setup_link() pad-op is taken before calling the op, while in other
scenarios the graph_mutex is taken after the isp->mutex.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Call media_pipeline_alloc_start() in stream start

Call media_pipeline_alloc_start() from atomisp_start_streaming() to mark
all involved subdevs as busy so that the links cannot be changed through
the media-controller APIs while streaming.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Propagate v4l2_mbus_framefmt.field to CSI port sink pad

Most sensor drivers report V4L2_FIELD_NONE as v4l2_mbus_framefmt.field
value. Propagate this to the v4l2_mbus_framefmt for the CSI port to pass
the link validation done by media_pipeline_start().

Keep the v4l2_mbus_framefmt.field for the source pad of the CSI port
as V4L2_FIELD_ANY to match the ISP sink pad.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Propagate set_fmt on sensor results to CSI port

So far the CSI port sink and source pads have not had their fmt set at all,
leaving things at the default of SBGGR8_1X8/0x0.

This means that pipeline validation such as e.g. done when calling
media_pipeline_start() will fail since the links to / from the CSI
ports have different fmts on each end.

Store a pointer to the CSI port v4l2-subdev in struct atomisp_input_subdev,
and use this in atomisp_set_sensor_crop_and_fmt() to propagate the sensors
new fmt after a successful set_fmt to the CSI port it is connected too.

The input->csi_port pointer also allows simplifying atomisp_link_setup().

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Ensure CSI-receiver[x] -> ISP links correctly reflect current sensor

Add a new atomisp_setup_input_links() helper which ensures that
the CSI-receiver -> ISP link for input_curr is marked as enabled and
the other CSI-receiver -> ISP links are disabled.

And call this helper from atomisp_register_device_nodes() for the initial
setup and from atomisp_select_input() for runtime input_curr changes.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Simplify atomisp_s_input() input argument checking

Simplify the checking of the input argument to atomisp_s_input():
- Remove the check for input >= ATOM_ISP_MAX_INPUTS, input_cnt
  never is > ATOM_ISP_MAX_INPUTS so checking only that is sufficient
- Remove the unnecessary camera local variable
- Move the 2 invalid argument checks to above the atomisp_pipe_check()
  which checks if the pipe is busy or not

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Add atomisp_select_input() helper

When switching input/sensor the s_power() callback must be called
for old sensor drivers to power on the new sensor and power off
the previous sensor.

atomisp_s_input() already does this but atomisp_link_setup()
did not do this.

Add a new atomisp_select_input() helper which does this and use this
in both atomisp_s_input() and atomisp_link_setup() for consistent
behavior.

Also make atomisp_link_setup() turn the sensor back off when
a link gets disabled.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Turn on sensor power from atomisp_set_fmt()

ATM the atomisp driver does not call s_power() at all when no s_input
ioctl() is done by the application. This breaks older sensor drivers
which rely on s_power() for power-management.

Some drivers have worked around this, e.g. commit c5fafbadaeae ("media:
atomisp: gc0310: Power on sensor from set_fmt() callback") and
commit b3118a942c82 ("media: atomisp: ov2722: Power on sensor from
set_fmt() callback"), but this really should be fixed in the atomisp
driver itself, so that all old drivers can work.

A logical place to call s_power() would be from atomisp_start_streaming() /
atomisp_stop_streaming(). But some older drivers, e.g. the atomisp-ov2722
driver already write mode related registers on set_fmt() instead of waiting
on stream on. So the s_power(1) needs to happen at the first set_fmt().

Add an atomisp_s_sensor_power(..., 1) call just before calling set_fmt()
for this. If the power was already enabled through e.g. a s_input ioctl
atomisp_s_sensor_power() will skip calling the s_power() v4l2-subdev-op
a second time.

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Add atomisp_s_sensor_power() helper

Since the atomisp driver only supports 1 stream / only has 1 asd now,
there is no longer a need to track which stream owns the sensor.

So the asd owner-pointer of struct atomisp_input_subdev can be dropped,
replace this with a simple camera_on boolean and add a new
atomisp_s_sensor_power() helper which takes care of avoiding unbalanced
s_power() subdev calls as well as takes care of handling the special
-ENOIOCTLCMD return for subdevs which don't implement s_power().

Reviewed-by: Andy Shevchenko <andy@kernel.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Get rid of PCI device ID hack check

Always check exact PCI ID match and drop hack checks.

Link: https://lore.kernel.org/r/20240402204447.294280-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Replace open-coded i2c_find_device_by_fwnode()

gmin_i2c_dev_exists() is using open-coded variant of
i2c_find_device_by_fwnode(). Replace it with a corresponding call.

Link: https://lore.kernel.org/r/20240326202813.1425431-3-andriy.shevchenko@linux.intel.com
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Put PMIC device after getting its I²C address

We don't use the PMIC I²C client device after getting its address.
Drop the reference to it. We do not expect device to disappear
as it should be taken care by the OpRegion drivers.

Link: https://lore.kernel.org/r/20240326202813.1425431-2-andriy.shevchenko@linux.intel.com
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: atomisp: Remove unsused macros

Entire defs.h can be killed, so do that.
Also kill same macro definitions in hive_types.h

Link: https://lore.kernel.org/r/20240326190903.1422069-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: atomisp: Remove extra whitespace after opening parentheses

This patch makes the following changes:
 * Removes spurious whitespace after the opening parentheses as per code
   style guidelines

Link: https://lore.kernel.org/r/20240227163043.112162-2-bergh.jonathan@gmail.com
Signed-off-by: Jonathan Bergh <bergh.jonathan@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

media: staging: media: atomisp: Fix various multiline block comment formatting instances

This patch makes the following fixes:
 * Reformats a number of multiline block comments to ensure * and */ align
   correctly

Link: https://lore.kernel.org/r/20240227163043.112162-1-bergh.jonathan@gmail.com
Signed-off-by: Jonathan Bergh <bergh.jonathan@gmail.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>