Move the documentation back to the mojom file from the cpp file. While at it, move the documentation for IPAInterface::init() and IPAInterface::stop() to the IPA guide. While at it, update the todo comment in all of the mojom files accordingly. Signed-off-by: Paul Elder <paul.elder@ideasonboard.com> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Reviewed-by: Umang Jain <umang.jain@ideasonboard.com> [umang.jain@ideasonboard.com: Update todos] Signed-off-by: Umang Jain <umang.jain@ideasonboard.com> Reviewed-by: Paul Elder <paul.elder@ideasonboard.com>
297 lines
9.0 KiB
Plaintext
297 lines
9.0 KiB
Plaintext
/* SPDX-License-Identifier: LGPL-2.1-or-later */
|
|
|
|
module libcamera;
|
|
|
|
/**
|
|
* \file core_ipa_interface.h
|
|
* \brief libcamera structs for IPAs
|
|
*/
|
|
|
|
/*
|
|
* Things that can be defined here (and in other mojom files):
|
|
* - consts
|
|
* - enums
|
|
* - structs
|
|
*
|
|
* Attributes:
|
|
* - skipHeader - structs only, and only in core.mojom
|
|
* - designate that this struct shall not have a C++ header definition
|
|
* generated
|
|
* - skipSerdes - structs only, and only in core.mojom
|
|
* - designate that this struct shall not have a (de)serializer generated
|
|
* - all fields need a (de)serializer to be defined, either hand-written
|
|
* in ipa_data_serializer.h
|
|
* - hasFd - struct fields or empty structs only
|
|
* - designate that this field or empty struct contains a FileDescriptor
|
|
*
|
|
* Rules:
|
|
* - Any struct that is used in a struct definition in mojom must also be
|
|
* defined in mojom
|
|
* - If the struct has both a definition in a C++ header and a (de)serializer
|
|
* in ipa_data_serializer.h, then the struct shall be declared as empty,
|
|
* with both the [skipHeader] and [skipSerdes] attributes
|
|
* - If the struct only has a definition in a C++ header, but no
|
|
* (de)serializer, then the struct definition should have the [skipHeader]
|
|
* attribute
|
|
* - Nested structures (e.g. FrameBuffer::Plane) cannot be defined in mojom.
|
|
* - Avoid them, by defining them in a header in C++ and a (de)serializer in
|
|
* ipa_data_serializer.h
|
|
* - If a struct is in an array/map inside a struct, then the struct that is
|
|
* the member of the array/map does not need a mojom definition if it is
|
|
* defined in a C++ header.
|
|
* - This can be used to embed nested structures. The C++ double colon is
|
|
* replaced with a dot (e.g. FrameBuffer::Plane -> FrameBuffer.Plane)
|
|
* - The struct must still be defined in a header in C++ and a (de)serializer
|
|
* implemented in ipa_data_serializer.h, as it cannot be defined in mojom
|
|
* - [skipHeader] and [skipSerdes] only work here in core.mojom.
|
|
* - If a struct definition has [skipHeader], then the header where the
|
|
* struct is defined must be #included in ipa_interface.h
|
|
* - If a field in a struct has a FileDescriptor, but is not explicitly
|
|
* defined so in mojom, then the field must be marked with the [hasFd]
|
|
* attribute.
|
|
*
|
|
* \todo Generate documentation from Doxygen comments in .mojom files
|
|
* \todo Figure out how to keep the skipHeader structs in sync with their
|
|
* C++ definitions, and the skipSerdes structs in sync with their
|
|
* (de)serializers
|
|
*/
|
|
[skipSerdes, skipHeader] struct ControlInfoMap {};
|
|
[skipSerdes, skipHeader] struct ControlList {};
|
|
[skipSerdes, skipHeader] struct FileDescriptor {};
|
|
|
|
[skipHeader] struct Point {
|
|
int32 x;
|
|
int32 y;
|
|
};
|
|
|
|
[skipHeader] struct Size {
|
|
uint32 width;
|
|
uint32 height;
|
|
};
|
|
|
|
[skipHeader] struct SizeRange {
|
|
Size min;
|
|
Size max;
|
|
uint32 hStep;
|
|
uint32 vStep;
|
|
};
|
|
|
|
[skipHeader] struct Rectangle {
|
|
int32 x;
|
|
int32 y;
|
|
uint32 width;
|
|
uint32 height;
|
|
};
|
|
|
|
/**
|
|
* \struct IPACameraSensorInfo
|
|
* \brief Report the image sensor characteristics
|
|
*
|
|
* The structure reports image sensor characteristics used by IPA modules to
|
|
* tune their algorithms based on the image sensor model currently in use and
|
|
* its configuration.
|
|
*
|
|
* The reported information describes the sensor's intrinsics characteristics,
|
|
* such as its pixel array size and the sensor model name, as well as
|
|
* information relative to the currently configured mode, such as the produced
|
|
* image size and the bit depth of the requested image format.
|
|
*
|
|
* Instances of this structure are meant to be assembled by the CameraSensor
|
|
* class by inspecting the sensor static properties as well as information
|
|
* relative to the current configuration.
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::model
|
|
* \brief The image sensor model name
|
|
*
|
|
* The sensor model name is a free-formed string that uniquely identifies the
|
|
* sensor model.
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::bitsPerPixel
|
|
* \brief The number of bits per pixel of the image format produced by the
|
|
* image sensor
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::activeAreaSize
|
|
* \brief The size of the pixel array active area of the sensor
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::analogCrop
|
|
* \brief The portion of the pixel array active area which is read-out and
|
|
* processed
|
|
*
|
|
* The analog crop rectangle top-left corner is defined as the displacement
|
|
* from the top-left corner of the pixel array active area. The rectangle
|
|
* horizontal and vertical sizes define the portion of the pixel array which
|
|
* is read-out and provided to the sensor's internal processing pipeline, before
|
|
* any pixel sub-sampling method, such as pixel binning, skipping and averaging
|
|
* take place.
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::outputSize
|
|
* \brief The size of the images produced by the camera sensor
|
|
*
|
|
* The output image size defines the horizontal and vertical sizes of the images
|
|
* produced by the image sensor. The output image size is defined as the end
|
|
* result of the sensor's internal image processing pipeline stages, applied on
|
|
* the pixel array portion defined by the analog crop rectangle. Each image
|
|
* processing stage that performs pixel sub-sampling techniques, such as pixel
|
|
* binning or skipping, or perform any additional digital scaling concur in the
|
|
* definition of the output image size.
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::pixelRate
|
|
* \brief The number of pixels produced in a second
|
|
*
|
|
* To obtain the read-out time in seconds of a full line:
|
|
*
|
|
* \verbatim
|
|
lineDuration(s) = lineLength(pixels) / pixelRate(pixels per second)
|
|
\endverbatim
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::lineLength
|
|
* \brief Total line length in pixels
|
|
*
|
|
* The total line length in pixel clock periods, including blanking.
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::minFrameLength
|
|
* \brief The minimum allowable frame length in units of lines
|
|
*
|
|
* The sensor frame length comprises of active output lines and blanking lines
|
|
* in a frame. The minimum frame length value dictates the minimum allowable
|
|
* frame duration of the sensor mode.
|
|
*
|
|
* To obtain the minimum frame duration:
|
|
*
|
|
* \verbatim
|
|
frameDuration(s) = minFrameLength(lines) * lineLength(pixels) / pixelRate(pixels per second)
|
|
\endverbatim
|
|
*/
|
|
|
|
/**
|
|
* \var IPACameraSensorInfo::maxFrameLength
|
|
* \brief The maximum allowable frame length in units of lines
|
|
*
|
|
* The sensor frame length comprises of active output lines and blanking lines
|
|
* in a frame. The maximum frame length value dictates the maximum allowable
|
|
* frame duration of the sensor mode.
|
|
*
|
|
* To obtain the maximum frame duration:
|
|
*
|
|
* \verbatim
|
|
frameDuration(s) = maxFrameLength(lines) * lineLength(pixels) / pixelRate(pixels per second)
|
|
\endverbatim
|
|
*/
|
|
struct IPACameraSensorInfo {
|
|
string model;
|
|
|
|
uint32 bitsPerPixel;
|
|
|
|
Size activeAreaSize;
|
|
Rectangle analogCrop;
|
|
Size outputSize;
|
|
|
|
uint64 pixelRate;
|
|
uint32 lineLength;
|
|
|
|
uint32 minFrameLength;
|
|
uint32 maxFrameLength;
|
|
};
|
|
|
|
/**
|
|
* \struct IPABuffer
|
|
* \brief Buffer information for the IPA interface
|
|
*
|
|
* The IPABuffer structure associates buffer memory with a unique ID. It is
|
|
* used to map buffers to the IPA with IPAInterface::mapBuffers(), after which
|
|
* buffers will be identified by their ID in the IPA interface.
|
|
*/
|
|
|
|
/**
|
|
* \var IPABuffer::id
|
|
* \brief The buffer unique ID
|
|
*
|
|
* Buffers mapped to the IPA are identified by numerical unique IDs. The IDs
|
|
* are chosen by the pipeline handler to fulfil the following constraints:
|
|
*
|
|
* - IDs shall be positive integers different than zero
|
|
* - IDs shall be unique among all mapped buffers
|
|
*
|
|
* When buffers are unmapped with IPAInterface::unmapBuffers() their IDs are
|
|
* freed and may be reused for new buffer mappings.
|
|
*/
|
|
|
|
/**
|
|
* \var IPABuffer::planes
|
|
* \brief The buffer planes description
|
|
*
|
|
* Stores the dmabuf handle and length for each plane of the buffer.
|
|
*/
|
|
struct IPABuffer {
|
|
uint32 id;
|
|
[hasFd] array<FrameBuffer.Plane> planes;
|
|
};
|
|
|
|
/**
|
|
* \struct IPASettings
|
|
* \brief IPA interface initialization settings
|
|
*
|
|
* The IPASettings structure stores data passed to the IPAInterface::init()
|
|
* function. The data contains settings that don't depend on a particular camera
|
|
* or pipeline configuration and are valid for the whole life time of the IPA
|
|
* interface.
|
|
*/
|
|
|
|
/**
|
|
* \var IPASettings::configurationFile
|
|
* \brief The name of the IPA configuration file
|
|
*
|
|
* This field may be an empty string if the IPA doesn't require a configuration
|
|
* file.
|
|
*/
|
|
|
|
/**
|
|
* \var IPASettings::sensorModel
|
|
* \brief The sensor model name
|
|
*
|
|
* Provides the sensor model name to the IPA.
|
|
*/
|
|
struct IPASettings {
|
|
string configurationFile;
|
|
string sensorModel;
|
|
};
|
|
|
|
/**
|
|
* \struct IPAStream
|
|
* \brief Stream configuration for the IPA interface
|
|
*
|
|
* The IPAStream structure stores stream configuration parameters needed by the
|
|
* IPAInterface::configure() method. It mirrors the StreamConfiguration class
|
|
* that is not suitable for this purpose due to not being serializable.
|
|
*/
|
|
|
|
/**
|
|
* \var IPAStream::pixelFormat
|
|
* \brief The stream pixel format
|
|
*/
|
|
|
|
/**
|
|
* \var IPAStream::size
|
|
* \brief The stream size in pixels
|
|
*/
|
|
struct IPAStream {
|
|
uint32 pixelFormat;
|
|
Size size;
|
|
};
|