The latest version of the Mali C55 driver has changed the format
accepted by the ISP sink pad to be the 20-bit wide.
The CSI-2 receiver handles the format expansion internally by
propagating the sensor produced format from its sink to the 20-bit
expanded version on its source pad.
Instead of re-applying the sensor format to the CSI-2 receiver source
pad (which is now an invalid operation) read from there the format
propagated by the driver internally and further propagate it to the ISP
subdevice.
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
Complete the RAW StreamConfiguration by populating the frame stride and
the frame size.
Set the minimum required alignment to 4 bytes as the Mali C55 ISP output
expands RAW output to 16 bits and a RAW Bayer macro-pixel requires
two samples to be complete.
Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
The PipelineHandler::completeBuffer() base class function returns
a boolean to indicate if there still are pending buffers in the
Request.
Simplify the bufferReady() function in the Mali-C55 pipeline handler
using the completeBuffer() return value.
Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The Mali C55 ISP cannot upscale. The maximum allowed output size
is the sensor's resolution.
For RAW streams this is already handled in adjustRawSizes(), while
for processed streams the maximum allowed resolution was wrongly
set to the ISP maximum output size (8192x8192).
Reviewed-by: Daniel Scally <dan.scally@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
The Matrix constructor that takes a std::vector is meant and only used
to initialize a Matrix from an initializer list. Using a std::vector is
problematic for two reasons. First, it requires constructing a vector,
copying the data from the initializer list, which is an expensive
operation. Then, the vector size can't be verified at compile time,
making the constructor unsafe.
The first issue could be solved by replacing the vector with a
std::initializer_list or a Span. The second issue would require checking
the initializer list size with a static assertion, or restricting usage
of the constructor to fixed-extent spans. Unfortunately, even if the
size of initializer lists is always known at compile time, the
std::initializer_list::size() function is a compile-time constant only
for constant initializer lists. Using a span would work better, but
construction of a fixed extent span from an initializer list must be
explicit, making the API cumbersome.
We can solve all those issues by passing an std::array to the
constructor. Construction of an array from an initializer list can be
implicit and doesn't involve a copy, and the array size is a template
parameter and therefore guaranteed to be a compile-time constant.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
In preparation to moving the matrix implementation from libipa to
libcamera copy the corresponding files to the new location. The files
are copied without modification to make upcoming integration changes
easier to see. The new files are not included in the build and therefore
have no negative side effects on the build.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
V4L2 Controls support a wide variety of types not yet supported by the
ControlValue type system.
Extend the libcamera ControlValue types to support an explicit 16 bit
unsigned integer type, and map that to the corresponding
V4L2_CTRL_TYPE_U16 type within the v4l2_device support class.
It's used on some camera metadata that is of length 16-bits,
for example JPEG metadata headers.
Signed-off-by: Yudhistira Erlandinata <yerlandinata@chromium.org>
Co-developed-by: Harvey Yang <chenghaoyang@chromium.org>
Signed-off-by: Harvey Yang <chenghaoyang@chromium.org>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The documentation seems to suggest that to create a new M2M
execution context it is expected users to call V4L2M2MDevice::open()
multiple times on the same video device path.
It is instead expected that multiple instances of the class are
created, one for each required execution context.
Clarify it in the documentation of the V4L2M2MDevice class.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
A copy is made in the range-based for loop, and thus all modifications
done in the for loop body are lost, and not actually applied to
the object in the container.
Fix that by taking a reference in the range-based for loop.
Fixes: 4217c9f1aa ("libcamera: camera: Zero streams before validate()")
Fixes: 613d540267 ("pipeline: raspberrypi: Fix handling of colour spaces")
Signed-off-by: Barnabás Pőcze <pobrn@protonmail.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Naushir Patuck <naush@raspberrypi.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add properties covering the sensor control application delays to both
the static CameraSensorProperties definitions. The values used are
taken from Raspberry Pi's CamHelper class definitions. Where no more
specific values are known the delay struct is defined as empty and
defaults supplied through the getter function.
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
The terms "shutter" and "shutter speed" are used through libcamera to
mean "exposure time". This is confusing, both due to "speed" being used
as "time" while it should be the inverse (i.e. a maximum speed should
correspond to the minimum time), and due to "shutter speed" and
"exposure time" being used in different places with the same meaning.
To improve clarity of the code base and the documentation, use "exposure
time" consistently to replace "shutter speed".
This rename highlighted another vocabulary issue in libcamera. The
ExposureModeHelper::splitExposure() function used to document that it
splits "exposure time into shutter time and gain". It has been reworded
to "split exposure into exposure time and gain". That is not entirely
satisfactory, as "exposure" has a defined meaning in photography (see
https://en.wikipedia.org/wiki/Exposure_(photography)) that is not
expressed as a duration. This issue if left to be addressed separately.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Naushir Patuck <naush@raspberrypi.com>
Acked-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The sensor is largely compatible with the already supported
Sony IMX290 so we can reuse the same helpers for the analogue
gain conversion functions.
Signed-off-by: Geoffrey Van Landeghem <geoffrey.vl@gmail.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
This patch introduces the configuration file for Virtual Pipeline
Handler. The config file is written in yaml, and the format is
documented in `README.md`.
The config file will define the camera with IDs, supported formats and
image sources, etc. In the default config file, only Test Patterns are
used. Developers can use real images loading if desired.
Signed-off-by: Konami Shu <konamiz@google.com>
Co-developed-by: Harvey Yang <chenghaoyang@chromium.org>
Signed-off-by: Harvey Yang <chenghaoyang@chromium.org>
Co-developed-by: Yunke Cao <yunkec@chromium.org>
Signed-off-by: Yunke Cao <yunkec@chromium.org>
Co-developed-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Add a test pattern generator class hierarchy for the Virtual
pipeline handler.
Implement two types of test patterns: color bars and diagonal lines
generator and use them in the Virtual pipeline handler.
A shifting mechanism is enabled. For each frame, the image is shifted to
the left by 1 pixel. It drops FPS though.
Add a dependency for libyuv to the build system to generate images
in NV12 format from the test pattern.
Signed-off-by: Konami Shu <konamiz@google.com>
Co-developed-by: Harvey Yang <chenghaoyang@chromium.org>
Signed-off-by: Harvey Yang <chenghaoyang@chromium.org>
Co-developed-by: Yunke Cao <yunkec@chromium.org>
Signed-off-by: Yunke Cao <yunkec@chromium.org>
Co-developed-by: Tomasz Figa <tfiga@chromium.org>
Signed-off-by: Tomasz Figa <tfiga@chromium.org>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The Fatal check of having at least one MediaDevice was to prevent
pipeline handler implementations searching and owning media devices with
custom conventions, instead of using the base function
|acquireMediaDevice|. It also has the assumption that there's at least
one media device to make a camera work.
Now that the assumption will be broken by the virtual pipeline handler
added in the following patches, and developers should be aware of the
available functions in the base class to handle media devices, the Fatal
check is no longer needed.
Signed-off-by: Harvey Yang <chenghaoyang@chromium.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
std::from_chars(), introduced in C++17, is a fast, locale-independent
string-to-arithmetic conversion function. The C++ standard library
provides overloads for all integer types, making it a prime candidate to
replace the manual handling of integer sizes in the YamlParser string to
integer conversion.
Compared to std::strtol(), std::from_chars() doesn't recognize the '0x'
prefix or '+' prefix, and doesn't ignore leading white space. As the
YamlParser doesn't require those features, std::from_chars() can be used
safely, reducing the amount of code.
C++17 also requires the standard C++ library to provide overloads for
floating-point types, but libc++ does not implement those. The float and
bool implementations of YamlParser::Getter::get() are therefore kept
as-is.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Debug metadata often occurs in places where the metadata control list is
not available e.g. in queueRequest() or processStatsBuffer() or even in
a class far away from the metadata handling code. It is therefore
difficult to add debug metadata without adding lots of boilerplate
code. This can be mitigated by recording the metadata and forwarding it
to the metadata control list when it becomes available. To solve the
issue of code that is far away from the metadata context, add a chaining
mechanism to allow loose coupling at runtime.
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
If the CameraManager fails to initialise at startup during
CameraManager::Private::init(), then the run() function will prepare the
thread for operations, but the thread will immediately close without
executing any cleanup.
This can leave instantiated objects such as the EventNotifier registered
by the udev enumerator constructed in a thread which no longer exists.
The destructor of those objects can then fire an assertion that they are
being executed from an incorrect thread while performing their cleanup.
Ensure that the failure path does not result in reporting thread
ownership assertions by performing cleanup correctly on the
CameraManager thread before it closes.
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
The names used by the IPA interface and the names used for buffer
completions handlers in libcamera clash in the use of the term "buffer".
For example video device buffer completion handler is called
"bufferReady" and the IPA event to ask the IPA to compute parameters are
called "fillParamsBuffers". This makes it hard to recognize which
function handles video device completion signals and which ones handle
the IPA interface events.
Rationalize the naming scheme in the IPA interface function and events
and the signal handlers in the pipelines, according to the
following table. Remove the name "buffer" from the IPA interface events
and events handler and reserve it for the buffer completion handlers.
Rename the IPA interface events and function to use the 'params' and
'stats' names as well.
IPA Interface:
- fillParamsBuffer -> computeParams [FUNCTION]
- processStatsBuffer -> processStats [FUNCTION]
- paramFilled -> paramsComputed [EVENT]
Pipeline handler:
- bufferReady -> videoBufferReady [BUFFER HANDLER]
- paramReady -> paramBufferReady [BUFFER HANDLER]
- statReady -> statBufferReady [BUFFER HANDLER]
- paramFilled -> paramsComputed [IPA EVENT HANDLER]
Cosmetic change only, no functional changes intended.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
PipelineHandler::stop() calls stopDevice() method to perform pipeline
specific cleanup and then completes waiting requests. If any queued
requests remain, an assertion error is raised.
Software ISP stores request buffers in
SimpleCameraData::conversionQueue_ and queues them as V4L2 signals
bufferReady. stopDevice() cleanup forgets to clean up the buffers and
their requests from conversionQueue_, possibly resulting in the
assertion error. This patch fixes the omission.
The problem wasn't very visible when
SimplePipelineHandler::kNumInternalBuffers (the number of buffers
allocated in V4L2) was equal to the number of buffers exported from
software ISP. But when the number of the exported buffers was increased
by one in commit abe2ec64f9, the assertion
error started pop up in some environments. Increasing the number of the
buffers much more, e.g. to 9, makes the problem very reproducible.
Each pipeline uses its own mechanism to track the requests to clean up
and it can't be excluded that similar omissions are present in other
places. But there is no obvious way to make a common cleanup for all
the pipelines (except for doing it instead of raising the assertion
error, which is probably undesirable, in order not to hide incomplete
pipeline specific cleanups).
Bug: https://bugs.libcamera.org/show_bug.cgi?id=234
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Tested-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Simple pipeline retrieves the requests to complete from the
conversionQueue_. This patch stores the requests in conversionQueue_
explicitly. This explicit tracking is supposed to be preferred to
implicit retrieval and it simplifies the completion code a bit here and
in the followup patch that adds request cleanup on stop.
The change as implemented assumes that all the buffers in each of the
conversionQueue_ elements point to the same request, the one specified.
Signed-off-by: Milan Zamazal <mzamazal@redhat.com>
Tested-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Tested-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The "useDewarper_" class variable is set to true during configure
only if the configuration is !isRaw.
The main path buffer pool creation can thus be moved out of the
if (!isRaw) block at allocateBuffers() time.
Cosmetic change that will make it easier to create a buffer pool
for the main and self paths unconditionally in future.
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Umang Jain <umang.jain@ideasonboard.com>
In order to support a default implementation for camera sensors when no
better implementation matches, libcamera needs to try "specialized"
implementations first and pick the default last. Make this possible by
adding a priority value for factories. Newly registered factories are
inserted in the factories list sorted by descending priority, and the
default factory uses a negative priority to be inserted as the last
element.
This mechanism may be a bit overkill in the sense that there is no
expected use cases for priorities other than trying the default last,
but the implementation is simple and easy to understand.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
With a camera sensor factory in place, the next step is to create an
abstract base class that all camera sensors implement, providing a
uniform API to pipeline handler. Turn all public functions of the
CameraSensor class into pure virtual functions, and move the
implementation to the CameraSensorLegacy class.
Part of the code is likely worth keeping as common helpers in a base
class. However, to follow the principle of not designing helpers with a
single user, this commit moves the whole implementation. Common helpers
will be introduced later, along with other CameraSensor subclasses.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Stefan Klug <stefan.klug@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Handle multiple scaler crops being set through the rpi::ScalerCrops
control. We now populate the cropParams_ map in the loop where we handle
the output stream configuration items. The key of this map is the index
of the stream configuration structure set by the application. This will
also be the same index used to specify the crop rectangles through the
ScalerCrops control.
CameraData::applyScalerCrop() has been adapted to look at either
controls::ScalerCrop or controls::rpi::ScalerCrops. The latter takes
priority over the former. If only controls::ScalerCrop is provided, the
pipeline handler will apply the same scaler crop to all output streams.
Finally return all crops through the same ScalerCrops control via
request metadata. The first configure stream's crop rectangle is also
returned via the ScalerCrop control in the request metadata.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Add a ispIndex field to CropParams that is used to track
which ISP output (0/1) will be used for a given stream during
configuration.
Tracking this information is required for an upcoming change where crop
rectangles can be specified for each configured stream. Currently, the
value is fixed to 0.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
In preparation for assigning separate crop windows for each stream, add
a new CropParams structure that stores the existing ispCrop_ and
ispMinCropSize_ as fields. Use a new std::map to store a CropParams
structure where the map key is the index of the stream configuration in
the CameraConfiguration vector.
At preset, only a single CropParams structure will be set at key == 0 to
preserve the existing crop handling logic.
Signed-off-by: Naushir Patuck <naush@raspberrypi.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
