It's confusing to read the code and understand that a request is only
completed before being processed by the ImgU if it only contains a
single RAW buffer. Add a boolean variable with a explanatory name to
make this clearer, no functional change.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Sharing a pointer with the IPA is fragile as it will not work with all
IPC mechanisms. Document that it's a temporary workaround that should be
reworked.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The generateConfiguration() implementation does not check if the
requested list of roles can actually be satisfied. The camera API
documentation prescribes the function shall fail in that case, instead
of silently adjust the returned configuration.
Fix this by implementing the same logic as the validate() function
implements, as the pipeline handler supports one raw stream and up to
two output streams.
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
The IPU3 pipeline handler that does not support receiving an empty list
of roles at generateConfiguration() time. This contradicts the camera
API which allows application to generate empty CameraConfiguration to
be later manually filled.
Fix this by returning an empty CameraConfiguration if the list of
requested roles is empty. While at it, align the style with the other
pipeline handlers.
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Jacopo Mondi <jacopo@jmondi.org>
The struct ImgUOutput now only contains one member that is in use, the
video device. Remove the struct and use the video device directly
instead.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The active_ flag is only used inside one function, remove the global
flag and handle it inside the single function.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Do not keep the duplicated ImgUDevice::ImgUOutput information in both
the stream and camera data classes. Remove it from the stream and only
access it from the camera data class.
Which stream is which can instead be checked by comparing it to the
known streams in camera data. This match how streams are checked in
other parts of the code making the pipeline more coherent.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
When the IPU3 pipeline only provided streams to applications that came
from the ImgU it made sense to have a generic function to configure all
the different outputs. With the addition of the RAW stream this begins
to be cumbersome to read and make sense of in the PipelineHandlerIPU3
code. Replace the generic function that takes a specific argument for
which sink to configure with a specific function for each sink.
This makes the code easier to follow as it's always clear which of the
ImgU sinks are being configured without knowing the content of a
generically named variable. It also paves the way for future
improvements.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
When the IPU3 pipeline was first developed sinks of the ImgU that where
not active still needed to have buffers allocated to allow streaming to
start. This is no longer true, it's enough that the sinks have imported
buffers to allow streaming to start. As we already need to import
buffers for stream that are active we can align the two cases and always
import buffers.
With this there is no longer a reason to store the allocated
FrameBuffers to keep them alive and the vector tracking them can be
removed.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Statistics buffers are not yet used by the IPU3 pipeline, they are never
queued to the statistics video device or in any other way consumed. The
kernel driver will however not allow video streaming to start if buffers
are not either allocated or imported on the statistics video device.
Instead of allocating the buffers wasting memory that is never used,
import buffers.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
With the refactored CIO2 interface it's now easy to add zero-copy for
buffers in the RAW stream. Use the internally allocated buffers inside
the CIO2Device if no buffer for the RAW stream is provided by the
application, or use the application-provided buffer if any.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The allocation and freeing of buffers for the CIO2 is handled in the
IPU3 pipeline handlers start() and stop() functions. These functions
also call CIO2Device start() and stop() at the appropriate times so
move the CIO2 buffer allocation/freeing inside the CIO2Device and reduce
the complexity of the exposed interface.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
In order to make the CIO2 easier to extend with new features make the
V4L2 devices (sensor, CIO2 and video device) private members. This
requires a few helper functions to be added to allow for the IPU3 driver
to still be able to interact with all parts of the CIO2. These helper
functions will later be extended to add new features to the IPU3
pipeline.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Collect the code used to generate configurations for the CIO2 block in
the CIO2Device class. This allows simplifying the code and allow further
changes to only happen at one code location.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Instead of spreading the mapping between media bus codes and V4L2 FourCC
all over the CIO2 code collect it in a single map and extract the data
from it. This is done in preparation of adding PixelFormat information
to the mix.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Decouple the number of buffers to allocate for the ImgU from the number
of buffers allocated for the CIO2. Instead of blindly following the CIO2
pick the maximum number of buffers requested for any stream facing
applications.
This is potentially wasteful, as each stream could allocate just as many
buffers as requested by the application instead of the maximum from the
set. But this is not more wasteful than what is already used by the
pipeline and should be fixed on top after the decoupling of the two
processing units.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Selecting which stream is the most suitable for the requested
configuration is mixed with adjusting the requested format when
validating configurations. This is hard to read and got worse when
support for Bayer formats was added. Break it out to a separate
function.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
The pipeline handler documentation incorrectly references an old API
usage of setCameraData, which should have been updated to
registerCamera() while updating pipeline handlers to ensure they all
have a pipeline-specific "CameraData" allocation.
Update the remaining documentation reference.
Fixes: b581b9576a ("libcamera: pipeline_handler: Make pipeline-specific data mandatory")
Signed-off-by: Chris Chinchilla <chris@gregariousmammal.com>
Reviewed-by: Umang Jain <email@uajain.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Print format names defined in formats namespace instead of the hex
values in toString() as they are easier to comprehend. For this add
a property of 'name' in PixelFormatInfo' so as to map the formats
with their names. Print fourcc for formats which are not used in
libcamera.
Signed-off-by: Kaaira Gupta <kgupta@es.iitr.ac.in>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
When replacing DRM FourCCs with libcamera formats IPU3 format
descriptors where missing resulting in RAW capture not being possible.
Fix this by adding the missing descriptors.
Fixes: 56c99424ed ("libcamera: pipeline: Replace explicit DRM FourCCs with libcamera formats")
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The id in the camera name is confusing and is of little use for users.
Camera names are not (yet) required to be unique and appending which
numerical CIO2 unit the sensor is attached to is just as good as
depending on the i2c bus information already present in the entity name.
Before this change,
$ cam -l
Available cameras:
1: ov13858 2-0010 0
2: ov5670 4-0036 1
After this change,
$ cam -l
Available cameras:
1: ov13858 2-0010
2: ov5670 4-0036
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The two public unsigned integer fields of StreamConfiguration, stride
and bufferCount where not initialized, fix this to match other members
being initialized to their 'zero' state.
Signed-off-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Jacopo Mondi <jacopo@jmondi.org>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Use the new pixel format constants to replace usage of macros from
drm_fourcc.h.
The IPU3 pipeline handler still uses DRM FourCCs for IPU3-specific
formats that are not defined in the libcamera public API.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
libcamera uses pixel format FourCC and modifier values from DRM. This
requires inclusion of drm_fourcc.h, creating a dependency on a header
that is packaged differently between distributions, and causing possible
issues with third-party applications.
Define constants for the supported pixel formats in the new formats.h
public API header, in order to remove the dependency on drm_fourcc.h.
The header is generated by a Python script from a list of supported
formats. The numerical values for the FourCC and modifier are extracted
from drm_fourcc.h by the script, ensuring that numerical values are not
inadvertently modified and preserving the direct interoperability.
The pixel formats constants can't be generated solely from drm_fourcc.h,
as that header defines FourCC values and modifier values, but doesn't
list the valid combinations. The supported formats are thus stored in a
YAML file, which contains the FourCC and optional modifier for each
supported format. We may later extend the YAML file to include formats
documentation, and possibly formats metadata to populate the
pixelFormatInfo map (in formats.cpp) automatically.
Now that two formats.h header are present (one in include/libcamera/ and
one in include/libcamera/internal/), we need to explicitly qualify the
Doxygen \file directive with a path.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund@ragnatech.se>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Emit 'cameraAdded' and 'cameraRemoved' from CameraManager to enable
hotplug and hot-unplug support in application like QCam.
To avoid use-after-free race between the CameraManager and the
application, emit the 'cameraRemoved' with the shared_ptr version
of <Camera *>. This requires to change the function signature of
CameraManager::removeCamera() API.
Also, until now, CameraManager::Private::addCamera() transfers the
entire ownership of camera shared_ptr to CameraManager using
std::move(). This patch changes the signature of Private::addCamera to
accept pass-by-value camera parameter. It is done to make it clear from
the caller point of view that the pointer within the caller will still
be valid after this function returns. With this change in, we can emit
the camera pointer via 'cameraAdded' signal without hitting a segfault.
Signed-off-by: Umang Jain <email@uajain.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Emit a signal whenever new MediaDevices are added to the
DeviceEnumerator. This will allow CameraManager to be notified
about the new devices and it can re-emumerate all the devices
currently present on the system.
Device enumeration by the CameraManger is an expensive operation hence,
we want one signal emission per 'x' milliseconds to notify multiple
devices additions as a single batch, by the DeviceEnumerator.
Add a \todo to investigate the support for that.
Signed-off-by: Umang Jain <email@uajain.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
This commit introduces no functional changes.
Split pipelines creation code into a separate function,
so that the function can be re-used for upcoming hotplug
functionality in subsequent commits.
Also, fixup correct tag for \todo.
Signed-off-by: Umang Jain <email@uajain.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The pipes_ vector was initially used to store pipeline handlers
instances with the CameraManager when it cannot be referenced from
anywhere else. It was used to retrieve cameras and deleting pipeline
handlers when stopping the camera manager.
In f3695e9b09 ("libcamera: camera_manager: Register cameras with the
camera manager"), cameras started to get registered directly with camera
manager and in 5b02e03199 ("libcamera: camera: Associate cameras with
their pipeline handler") pipeline handlers started to get stored in a
std::shared_ptr<> with each camera starting to hold a strong reference
to its associated pipeline-handler. At this point, both the camera
manager and the camera held a strong reference to the pipeline handler.
Since the additional reference held by the camera manager gets released
only on cleanup(), this lurking reference held on pipeline handler did
not allow it to get destroyed even when cameras instances have been
destroyed. This situation of having a pipeline handler instance around
without having a camera may lead to problems (one of them explained
below) especially when the camera manager is still running.
It was noticed that, there was a dangling driver directory issue (tested
for UVC camera - in /sys/bus/usb/drivers/uvcvideo) on 'unbind' → 'bind'
operation while the CameraManager is running. The directories were still
kept around even after 'unbind' because of the lurking reference of
pipeline handler holding onto them. That reference would clear if and
only if the CameraManager is stopped and then only directories were
getting removed in the above stated path.
Rather than writing a fix to release the pipeline handlers' reference
from camera manager on camera disconnection, it is decided to eliminate
the pipes_ vector from CameraManager moving forwards. There is no
point in holding a reference to it from camera manager's point-of-view
at this stage. It also helps us to fix the issue as explained above.
Now that the pipeline handler instances are referenced via cameras only,
it can happen that the destruction of last the camera instance may
result in destruction of the pipeline handler itself. Such a possibility
exists in PipelineHandler::disconnect(), where the pipeline handler
itself can get destroyed while removing the camera. This is acceptable
as long as we make sure that there is no access of pipeline handler's
members later on in the code path. Address this situation and also add a
detailed comment about it.
Signed-off-by: Umang Jain <email@uajain.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
